Investigating Citizen Perceptions and Business Performance of Airbnb in Korea

Abstract

The "sharing economy" describes a type of business built on the sharing of resources - allowing customers to access goods when needed. While sharing goods has always been a common practice among friends, family, and neighbors, in recent years, the concept of sharing has moved from a community practice into a profitable business model. This study explores gaps between customers' actual usages and current policies on accommodation sharing by analyzing what needs to be done for the better establishment of sharing economy in society. The purpose of this study is to investigate perceptions of accommodation sharing by analyzing reviews and negative aspects that help resolve complaints and improve better services through policy establishment. This study investigates key attributes that influence business performance to improve citizens' decision-making for the usage of accommodation sharing. This study applies qualitative research by collecting demand- and supply-side reviews from selected registered accommodations using a random sampling procedure. This study finds that guests prefer entire house sharing with instrumental attributes related to properties. Entire house sharing of multiple dwellings shows business impacts in terms of high occupancy rate on the platform, while there are policy concerns with entire house sharing. The results provide policy and managerial implications by suggesting proper policies and considering relationships with citizens.

1. Introduction

 The rapid growth of the sharing economy or collaborative consumption has rapidly changed the trends of consumers, which has sparked mounting interest from both academia and industry professionals. The sharing economy is a peer- to-peer activity that allows people to share or obtain goods and services, often through online platforms. In the tourism and hospitality sector, Airbnb is a prominent example of peer-to-peer accommodation, enabling people to offer their unoccupied houses or rooms for short-term rental (Sundararajan, 2014) at low searching and operation costs (Henten & Windekilde, 2016). Over the last decade, the rise of the sharing economy as “consumers granting each other temporary access to under-utilized assets (‘idle capacity’), possibly for money”—has received growing attention because of its potential to foster more sustainable lifestyles (Richardson, 2015; Van Dijck, 2009). Many studies that define the modern concepts of sharing economy, use various terminologies such as the access-based economy (Bardhi & Eckhardt, 2012), collaborative economy (Botsman, 2015), and on-demand economy (Jaconi, 2014). This study defines sharing economy as - a socio-economic system built around the sharing of resources. It often involves a way of purchasing goods and services that differs from the traditional business model of companies hiring employees to produce products to sell to consumers. It includes the shared creation, production, distribution, trade, and consumption of goods and services by different people and organizations. These systems take a variety of forms, often leveraging information technology to empower individuals, corporations, non- profits, and governments with information that enables distribution, sharing, and reuse of excess capacity in goods and services. Lessig (2008) described the sharing economy as being regulated by a set of social relations as opposed to the commercial economy where access is based only on the metric of price (Cohen & Kietzmann, 2014; Zervas et al., 2017). The sharing economy is expected to grow from $14 billion in 2014 to $335 billion by 2025 (Yaraghi & Ravi, 2016). The Bank of Korea (2019) estimates that the digital sharing economy between peers grew tenfold from 20 billion won (17 million dollars) in 2015 to 198 billion Won (168 million dollars) in 2018.

 The rapid growth of peer-to-peer accommodation such as Airbnb can be partly attributed to its offering an alternative accommodation experience for guests. Peer-to- peer accommodation provides guests a local experience by enabling guests to interact with the hosts or neighbors. This study investigates accommodation sharing; accommodation sharing provides more accessible and affordable hospitality services with lower costs. Airbnb has started the peer-to- peer accommodation sharing service in Korea in 2014. In 2018, more than 70, 000 accommodation sharing has been registered in Korea. The Airbnb historical data in Korea is provided by AirDNA, which collects publicly available Airbnb data of various cities from the Airbnb website and provides a market analysis report. Among regions, Seoul has 26, 516 registered accommodation sharing which is 36% of total accommodation sharing. Jeju is the second largest accommodation sharing including 11, 502 listings (16%), followed by Gangwon, Busan, Gyeonggi, and other provinces. The aggregate number of accommodations in Seoul, Gyeonggi, Gangwon, and Jeju is 70% of the total listed accommodation and shows apparent concentrations in several cities and provinces. By applying the geographical coordinates of accommodations, the geographical distribution and the density of accommodation sharing registered on the Airbnb platform can be visualized. This study presents the density of accommodation sharing in Figure 1 by applying the optimized hot spot analysis with point features, which calculates the Getis-Ord Gi* statistic (pronounced G-i-star) for each feature in a dataset. The resultant z-scores and p-values tell you where features with either high or low values cluster spatially. These are statistically significant spatial clusters of high values by ArcGIS, a location-based analytical software. ArcGIS provides contextual tools and services for mapping and spatial analysis so you can explore data & share location-based insights.

 By considering the volume and density of accommodation sharing supplier sides, this study examines the case of Seoul, where the density is high among all other cities in Korea. As shown in Figure 1, Seoul has five districts with a high level of density as these districts are widely commercialized compared to other districts and located relatively in the middle of the city along the Han River. This study focuses on registered accommodations to evaluate the impacts of accommodation sharing in urban and large cities.

Figure 1: The Density of Accommodation Sharing in Korea and Seoul information system with the longitude and latitude

 Huh and Noh (2018) analyzed the locational characteristics of Airbnb in Seoul and identified potential problems in urban planning. First, they analyzed the operation status of Airbnb in terms of the commercial host. After that, they identified spatial distributions of Airbnb and examined influential factors that affected the locational characteristics of Airbnb. The results of this study Airbnb units were mainly located in residential areas, especially around multiplexes and townhouses

 Kim et al. (2017). This study utilizes the number of reviews on the Airbnb platform as to how experienced guests evaluate P2P accommodations. This study applies qualitative research by analyzing guests’ reviews and hosts’ responses from Airbnb to identify the main characteristics. The negative factors of accommodation sharing are drawn from the guests’ complaints. The positive factors can foster accommodation sharing, while negative factors may cause guests to hesitate to use the sharing.

 The purpose of the first study is to explore how customers perceive factors of accommodation sharing (Study 1). Further, this study investigates negative aspects of accommodation sharing based on perspectives of the demand side. The results of Study 1 provide implications on policy reactions that might require proper regulations to improve reliability and better usages of accommodation sharing. Second, this study investigates the effects of attributes on the performance of accommodation sharing by using quantitative secondary data and draws implications on specific preferences of experienced guests (Study 2). While hotels provide standardized rooms and services to guests, accommodation sharing provides diverse types of accommodations in terms of properties owned by individuals. Study 2 explores how the factors of accommodation sharing including price, size of space, type of property, availability of amenities affect business performance in terms of occupancy rate.

2. Study 1: Analysis of Perceptions

2.1. Methodology

 This study applies content analysis to understand the significant attributes from reviews of accommodation sharing and to identify the main attributes of accommodation sharing. Content analysis is a research technique used to make replicable and valid inferences by interpreting and coding textual material. By systematically evaluating texts (e.g., documents, oral communication, and graphics), qualitative data can be converted into quantitative data. This study uses both Leximancer and Nvivo software packages to identify the main concepts and themes by analyzing word frequency from the reviews. By applying for guests’ reviews, this study finds that the guests have strong impressions in terms of property, hosts, and other services related to accommodation sharing. Sentiment Analysis is the process of ‘computationally’ determining whether a piece of writing is positive, negative, or neutral. It’s also known as opinion mining, deriving the opinion or attitude of a speaker. This is applied to this study to be aware of the proportion of positive and negative reviews, to find risk factors of accommodation sharing, and to classify the type of negative sentences in the attributes of accommodation sharing. This study applies guests’ reviews of Airbnb to investigate key characteristics of accommodation sharing during experiences and significant factors, so as to evaluate provided services. To extract samples from a volume of reviews on the platform, this study applies random sampling procedures by IDs and procedures for the selection and analysis of guests’ reviews and hosts’ responses (Figure 2).

Figure 2: Review Selection and Analysis Procedure

 This study selected reviews based on the listings of AirDNA that have listings of Airbnb data. From the listings of AirDNA, the study randomly selected 600 samples of accommodation sharing and collected the reviews. This study selected the reviews based on sharing types with residential properties and focuses on the 8, 385 guest reviews and 958 host reviews.

2.2. Analysis of Reviews

 This study finds major factors of accommodation sharing by analyzing the guests’ reviews and corresponding hosts’ responses. By classifying the reviews, this study improves the thematic comprehension of clustering words of like meaning. The concepts of guests’ reviews were clustered based on the level of themes, so 73 concepts were classified into six themes: location, experience, sanitation, property, host, and appreciation, and they are visualized in Figure 3.

Figure 3: Themes of Airbnb Guests’ Reviews of All Types

*Leximancer Outcome: Visual Concept 100%, Theme Size 50%, Otherwise the Default Settings.

 Major clusters include location, experience, sanitation, property, host, and appreciation related; i) location is associated with physical location and transportation such as subway station, convenience, walking distance, shopping and food, bus, street, and more; ii) experience is associated with stay, during the trip, price, best, wonderful, experience, and visit; iii) sanitation is associated with clean, comfortable, cozy, space, while it shared with concepts including properties and facilities such as a house, room, apartment, kitchen, bed, and bathroom; iv) property is associated with the accessibility of facilities such as a bathroom, bed, provided, floor, kitchen, and water; v) host is associated with hospitality which is provided to guests; and vi) appreciation is associated with guests’ feelings of gratitude, mostly to hosts and their hospitality. The concepts of hosts’ responses are clustered into five themes: appreciation with expression, experience, evaluation, explanation, and expectation.

 The concepts of hosts’ responses are clustered into five themes: appreciation, experience, evaluation, explanation, and expectation; i) appreciation is an expression of hosts’ feeling of gratitude for staying with them or leaving good reviews; ii) experience is associated with personal interaction between hosts and guests such as time, trip, family, enjoy, memories and others; iii) evaluation is associated with guests; iv) explanation is an opportunity for hosts to justify guests’ dissatisfaction or complaints; and v) expectation is associated with willingness to host the guests in the future. The results of analyzing both the conceptual map and the classification of key concepts based on word frequency study finds that guest reviews are related to behavioral and emotional measurements and physical features of accommodation sharing, while the hosts’ responses mainly focus on the behavioral and emotional contents.

2.3. Analysis of Negative Reviews

 This study investigates the selected negative reviews to examine the demand side perceptions. First, this study examines how accommodation sharing types and property types are related to the listed negative factors. This study investigates accommodation sharing types and property types since many policy concerns are related to types of property and sharing. Sharing the entire house is strictly prohibited in cities such as New York and the majority of cities in Korea. This study hypothesized that there is a relationship between sharing types and complaints as guests tend to prefer entire house sharing rather than other types (based on reviews). This study also hypothesized that residential property types are related to the types of complaints.

H₁: Accommodation sharing types (entire house sharing, private room sharing, or shared room sharing) are not independent of types of complaints.

H₂: The housing (residential property such as apartments or houses) types are not independent of the types of complaints.

 Further, this study hypothesized that complaints are associated with the reliability of hosts. Airbnb awards superhost badges to hosts who have a higher average overall rating based on the reviews. Therefore, the superhost badge is a signal of outstanding quality and could hence help to build the reputation of the host (Teubner et al., 2017).

H₃: The status of being a superhost is not independent of the types of complaints.

 Accommodation sharing data shows that registered accommodations are usually located in highly dense areas, and complaints and evaluations are associated with sanitation, location, hosts, and properties based on the analysis of guests’ reviews (Dudás et al., 2017; Ki & Lee, 2019). This study examines the relationship between geographical distribution and the types of complaints. The unbalanced distribution might require specified policies and regulations in each district.

H₄: The districts of accommodation sharing are not independent of the types of complaints.

2.4. Results of Negative Reviews

 To test the hypothesis, this study applies sentiment analysis and conceptual analysis. This study uses 629 negative reviews for the negative sentiment analysis. These reviews are categorized into seven complaints by manually reviewing each negative references, regardless of the degree of dissatisfaction: accuracy (29 counts, 5%), communication (33 counts, 5%), host (22 counts, 3%), location (160 counts, 25%), property (237 counts, 38%), security & safety (33 counts, 5%), and sanitation (115 counts, 18%). This study conducts a chi-square analysis to examine the relationship between accommodation sharing types and the types of complaints that are classified based on the results of the proportion (H₁). The result of H₁ showed significance at χ² = 34.198, P < 0.01 and confirms that there is a relationship between accommodation sharing types and the types of complaints. Chi-square analysis for residential property types such as apartments and houses and complaints (H₂) does not show significance at χ² = 5.370, P > 0.1. There is no relationship between residential property types and the types of complaints. In terms of host reliability (H₃), the result of chi-square shows significance (H₃) at χ² = 29.364, P < 0.01, so complaints are related to the reliability of hosts. Hosts with superhost badges deliver relatively better services and have fewer complaints or negative reviews. Location is one of the key attributes of accommodation sharing based on conceptual mapping and word frequency analysis. The result of chi-square accepts H₄ at χ² = 61.984, P < 0.01 and shows that the districts of accommodation sharing are related to the types of complaints.

3. Study 2: Analysis of Business Performance

3.1. Methodology

 Study 2 uses cross-sectional data of Airbnb from January 1 to December 31 in 2018, provided by AirDNA. The total number of listed accommodations in Seoul is 26, 516 registered accommodations. This study includes the listed accommodations with an occupancy rate greater than zero, so the total number of observations is 17, 065 registered accommodations. The data has been reclassified in terms of types of property, numbers of rooms, districts of accommodations, and others (Table 1).

Table 1: The Summary of Descriptive Statistics

Table 1: Continue

Note: Includes accommodation sharing with housing types (apartment, house, condominium, loft) and with occupancy rate (>0)

 The platform provides property information such as the number of rooms, property types, sharing types, and access to amenities. The original data consists of the entire house sharing (11, 441), private sharing (4, 836), and shared room sharing (768). The types of property are classified into four residential housing types including apartments (9, 072), houses (5, 097), condominiums (2, 103), and lofts (793). The data includes information related to the price and expenses of the accommodations, including the average daily rate (price) in US dollars, and additional fees such as deposit, cleaning charges, and extra guest fees. The average daily rate shows 90.63 dollars for entire room sharing, 43.22 dollars for private room sharing, and 27 dollars for shared rooms. This study deals with accommodation sharing with at least more than one reservation during the available booking days. The average occupancy rate is 56.69% for all sharing types, 63.03% for the entire house, 45.36 % for private rooms, and 33.40% for shared rooms. To find the relationship between attributes of accommodation sharing and performance, this study constructs multiple regression analysis.

3.2. Hypotheses Development

 Study 2 examines how critical attributes of accommodation sharing influence business performance. Study 2 attempts to develop the attributes from analyzing guests’ reviews and include the corresponding variables to Airbnb transactional data. The proposed attributes include property attributes such as sharing types, housing types, number of rooms, and availability of amenities, and pricing attributes related to economic benefits, location attributes, and host and communication attributes. Qiu et al. (2018) stated that property attributes become significant factors that influence the better performance of Airbnb.

H₁ – H₄: The property attributes (sharing types, housing types, number of rooms, and availability of amenities) significantly influence the performance of accommodation sharing.

 According to the previous study, economic benefits are a critical factor in sharing economy (Hamari et al., 2016). People expect to have a sharing economy service at a lower cost with various options (Schor, 2016). Gunter and Önder (2018) stated that hosts set their own prices for accommodation sharing and the market price for sharing is highly uncompetitive.

H₅ – H₆: The pricing attributes (average daily rate, the status of additional fees) significantly influence the performance of accommodation sharing.

 This study measures the impact of locational preference on the guests’ choices/preferences. The description of the location is available on the profile of each accommodation, so this study hypothesizes that the attributes related to the location center (Dogru & Pekin, 2017; Wang & Nicolau, 2017) influence performance.

H₇ – H₉: The location attributes (districts, accessibility of transportation, locational convenience) significantly influence the performance of accommodation sharing.

 In terms of the peer-to-peer sharing economy, the trust between providers such as hosts in accommodation sharing and the user such as guests becomes significantly important. The benefits from the interactions with hosts increase satisfaction (Tussyadiah, 2016). The host characteristics might also influence pricing and performance of accommodation sharing, as such, hosts with more accommodations and experiences expect different pricing and receive higher reservations (Guttentag & Smith, 2017; Tussyadiah, 2016). This study aims to measure the significance of hosts attributes such as multiple listings and superhost badges (Liang et al., 2017) since hosts’ efforts including updating photos and response toward guests as well as the reviews from the guests of their performance might affect the price of accommodation (Xie & Mao, 2017). The study also includes the option of instant booking as it might increase the convenience of accessibility to the shared accommodation.

H₁₀ – H₁₄: The host and communication attributes (the status of superhost, the status of multiple listing, number of reviews, response time and rate, instant booking status) significantly influence the performance of accommodation sharing.

 To measure how the evaluation by previously experienced guests influences the purchasing decision of future guests, this study hypothesizes that rating by previously experienced guests of the accommodation influences the number of nights guests stay at each accommodation. Airbnb requests experienced guests to evaluate the services of each registered accommodation, in terms of six aspects including accuracy, check-in, cleanliness, communication, location, and values. The ratings are shown on the platform with the overall rating.

H₁₅: The rating on accommodation sharing by experienced guests significantly influences the performance of accommodation sharing.

 This study investigates the impacts of the listed attributes on performance based on each accommodation sharing type because this study assumes that types of sharing (including entire houses, private rooms, and shared rooms) would have different influences on guests’ preferences. The listed hypotheses are demonstrated in Figure 4.

Figure 4: Framework for Impacts of Accommodation Sharing Attributes on Performance and Price

3.3. Data Analysis

 This study develops hypotheses for sharing types and property types (H₁–H₄), price (H₅–H₆), location (H₇–H₉), host and communication (H₁₀–H₁₄), and overall rating (H₁₅) and their impacts on business performance. This study uses occupancy rates as a signal to business performance. Model (1) includes all the listed attributes and model (2) is restricted to entire house sharing, model (3) is restricted to private room sharing, and model (4) is restricted to shared room sharing, because this study assumes that the impact of the attributes on the occupancy rate might be different in terms of accommodation sharing types (Table 2).

Table 2: The Results of Multivariate Regression of Performance (Occupancy Rate)

Table 2: Continued.

***Significant at 0.01 level (2-tailed); **Significant at 0.05 level (2-tailed); *Significant at 0.1 level (2-tailed), Robust Standard Error applied.

3.3.1. Analysis of Property Attributes

 Accommodation sharing types (H₁a–H₁b) are significant at α = 0.01 level in the model (1) and confirms that private room sharing (H₁a) and shared room sharing (H₁b) significantly decrease the occupancy rate of accommodation sharing, compared to the entire house sharing. In addition, the mean of occupancy rate of entire house sharing is 63%, while the mean of private room sharing is 45% and the mean of shared room sharing is only 33%. It means that guests have a strong preference for entire house sharing. This model includes four housing types including apartment, house, condominium, and loft. Among the housing types, house (H₂a) and condominium (H₂b) show significance at α = 0.01 level, and loft (H₂c) is significant at α = 0.05 level, for model (1). It means that house has a negative influence on the occupancy rate in comparison with sharing in apartments while sharing in condominium and loft have a relatively positive influence. The results find that property types are significant to the occupancy rate. According to the number of rooms, the study assumes that accommodation sharing with more than three rooms provides benefits for travelers with families and companions to stay at the same property. The study finds that the number of rooms from one to three rooms compared to studio types is insignificant, but the properties with more than three rooms (H₃d and H₃e) are significant at α = 0.01. It indicates that guests with travel companions prefer independent units/property with many rooms, but the number of properties with more than three rooms is only 392 registered accommodations (2.3%), particularly in the case of the entire room sharing. The study includes the influence of amenities such as kitchen, Internet, breakfast, laptop computer, and elevators (H₄a~H₄e), which are mentioned frequently in the guests’ reviews. The availability of these amenities is significant at α = 0.01 in models (1) and (2). Private room sharing shows that the availability of breakfast and computers are significant in model (3). The study finds that the availability of breakfast has a significantly negative impact on the occupancy rate.

3.3.2. Analysis of Value Attributes

 The study estimates the effects of the average daily rate (price) on occupancy rate and the average daily rate (H₅) is significant at α = 0.01. This study finds that price is a critical factor for guests to choose their accommodation. From the qualitative study, guests have not frequently mentioned the price attributes, but the quantitative study finds the impact of price factor on occupancy rate is significant and that guests consider price factor when choosing the accommodations. This study also tests whether the existence of additional charges such as security deposit, cleaning fee, and extra guest fees might impact the occupancy rate. The results show that the cleaning fee (H₆b) is positively significant and the extra guest fee (H₆c) is negatively significant at α = 0.01 in models (1) – (4). It can be presumed that guests might expect neat and organized accommodations in spite of separately charged cleaning fees, and cleanliness (i.e. sanitation) is shown as the second-largest cluster on the conceptual map of guests’ reviews (Figure 5). The extra guest fee (H₆c) is negatively significant at α = 0.01 in models (1) – (4). It explains that guests are aware of the capacity of accommodation and are unwilling to pay the extra guest fee if they can manage to stay together. Results show that security deposit is insignificant.

3.3.3. Analysis of Locational Attributes

 According to districts where registered accommodation sharing is located, accommodations in Yongsan (H₇b), Jongno (H₇d), and others (H₇e) are significant at α = 0.01 and it indicates that accommodations located in Yongsan, Jongno, and other districts are less preferred than accommodations in Mapo. In the case of private room sharing, Jongno and other districts are negatively significant in model (3), while the districts of shared rooms are insignificant in model (4). The density of accommodation sharing in each district and the occupancy rates necessitate that each municipality should have localized regulations and promotion strategies based on the density and business performance. The accessibility to accommodation sharing, the distance to the nearest subway station (H₈a) is significant at α = 0.01 in models (1) and (2) and α = 0.1 in model (4), while the distance to the nearest bus stop (H₈b) is insignificant. It explains that guests prefer easy access to the subway. Results show that the locations near café (H₈c) are negatively significant in the case of entire house sharing and the distance to convenience stores (H₈d) is insignificant. It means that easy access to these facilities provides convenience, but does not affect guests’ choices. The number of restaurants within 1 km (H₈e) is negatively significant in models (1), (3), and (4). However, the number of shopping sites (H₈f) and touristic sites are significant. The results posit that guests prefer accommodation sharing with easy accessibility and more facilities such as shopping and touristic sites, but they tend to avoid staying near areas with a higher number of restaurants and cafes.

3.3.4. Analysis of Host & Communication Attributes

 The host characteristics and communication are critical factors to build trust in the P2P sharing economy. This study includes the status of superhost (H₁₀a), the status of multiple accommodation hosting (H₁₀b), number of reviews (H₁₁) and photos (H₁₂), response rate (H₁₃a), and response time (H₁₃b), as well as the possibility of instant booking (H₁₄). All these listed factors are significant at α = 0.01 in model (1) and model (2). However, for private room sharing in model (3), super host, the number of reviews, and instant booking are significant at α = 0.01, and the response rate is significant α = 0.05. For shared room sharing in model (3) the number of reviews, photos, and instant booking, are significant at α = 0.01. The findings also provide some implications that the status of being superhost and receiving a greater number of reviews become signals of trust-building on peer-to- peer transactions (Teubner et al., 2017). It indicates that the number of reviews and the contents of the reviews influence guests’ decisions.

3.3.5. Analysis of Overall Rating

 The overall rating which is evaluated the accommodation by the experienced guests (H15) is significant at α = 0.01. As far as the platform provides the rating information, it would influence the guests’ choice and differ the occupancy rate.

4. Conclusion

4.1. Summary of Findings

 From the selected guests’ reviews and the corresponding hosts’ responses, Study 1 finds that guests’ reviews focus on behavioral and emotional measurements and physical features of accommodation sharing, while the hosts’ responses mainly focus on the behavioral and emotional contents. Guests frequently mention property, sanitation, location, and hosts, while hosts mention their appreciation toward guests and guests’ reviews and experiences, evaluation of the guests, and their expectations to host the guests. The negative reviews from the sentimental analysis are classified into seven categories such as accuracy, communication, host, location, property, sanitation, and security. These complaints are highly related to location, property, and sanitation for all accommodation sharing types. Based on the results of chi-square, the study finds that the categories of complaints are significantly related to accommodation sharing types, the status of superhost, and the location of accommodation, but are not related to the housing types whether the accommodation sharing is in a house or an apartment.

 Study 2 finds that each sharing type such as entire house sharing, private room sharing, and shared room sharing, and other attributes of accommodation sharing impact accommodation sharing. Entire house sharing is more popular with large number of accommodations registering on the platform and having a higher occupancy rate than private rooms or shared rooms. The small size of apartment, condominium, or loft is preferred, compared to a house. With relatively few numbers of properties with more than four rooms, the results show positive impacts on reservations in case of entire sharing. It infers that many guests have a strong preference for entire house sharing so as to have independent accommodation and privacy and to be able to stay with friends or families (especially where there are more than three members) in the same property, unlike the standard budget hotels. The price of accommodation becomes significant while cleaning expenses positively influence accommodation sharing. It shows that guests consider sanitation as a critical factor and this is consistent with the findings from the review analysis (guests frequently use sanitation-related words and a high proportion of the complaints concern sanitation). Accommodation sharing is concentrated in five districts in Seoul. Location in the center of the city with more convenient facilities such as café, shopping sites, and touristic sites are preferred by guests. Particularly, the distance to the subway station has more influence than the distance to the bus stop on accommodation sharing. Also, in terms of attributes related to hosts, the status of superhost badges and the number of reviews and photos are significant on guest preference and increase the occupancy rate. This information can improve trust in the peer-to-peer sharing economy. The response time and rate, available instance booking, and the overall rating by former guests improve the convenience of accommodation sharing.

4.2. Policy and Managerial Implication

 This study finds the necessity of policy and managerial reactions. Particularly, the findings based on individual perceptions from the negative reviews against security, hygiene, safety, information accuracy, communication, location, and property emphasize the importance of quality control to secure reliable accommodation sharing. Not only accommodation sharing but also the sharing economy needs proper legal backgrounds to protect users (guests) and service providers (hosts). Because sharing economy is a transaction between peers, the user should aware of the guidelines, while personal providers must protect business and personal assets. For protecting P2P sharing services, the roles and responsibilities of platform providers and governments become very critical. For efficient compliance, platform providers can use voluntary self-regulation to filter out illegal transactions and take insurance schemes.

 The government needs to apply policies such as regulatory-sandbox to resolve the regulatory problems. Under established laws and regulations to legalize sharing business, a ‘standards-compliance certificate’ must be made mandatory for all service providers to maintain the service quality and to increase the trust of guests and hosts in the sharing economy. Furthermore, sharing economy has considerable influence on communities, so the government needs to communicate with local residents to prevent illegal services and prevent damages to the community. Therefore, the government should establish strategically appropriate policies for the mass public. This study results show that citizens prefer entire house sharing without knowing fact that many forms of entire house sharing without hosts are considered illegal. Such issues could be mitigated by applying proper public promotions and advertisements, targeted to citizens. According to the business perspective, sharing economy can provide a wide range of business opportunities in terms of individual participation, business collaboration, and extension of global services. First, the platform can lead to the active participation of individuals for P2P business and create social and community benefits for fostering the local economy. Second, sharing economy can provide integrated services such as local accommodation and mobility sharing as well cultural experience by local residents. Last, sharing economy can extend its global services by connecting to citizens across countries; however, it is difficult to overcome the differences in housing and social culture and the legality of sharing economy.

4.3. Limitations

 In the case of qualitative research, the reviews strongly incline toward positive reviews compared to negative reviews, because customers have fewer complaints or dissatisfaction. Also, the analysis of word frequency has the possibility to exclude important variables due to a small number of frequencies. This study uses one-year data of Airbnb, hence, future studies should consider many years of data and also investigate seasonality. Further, impacts due to Covid-19 could be studied in the future. Further study might also consider other aspects such as the performance of the host (Nguyen & Ngo, 2020), the innovative approach of sharing economy (Nguyen, 2020), and policy issues (Dahliah et al., 2020).