Do airbnb host listing attributes influence room pricing homogenously?
Introduction
The global regression models used for investigating the influence of determinants on room pricing of Airbnb’s sharing accommodation mask the spatial heterogeneity of relations (Wang and Nicolau, 2017; Zhang et al., 2011). These models possibly overlook the presence of regional (or city-specific, in the case of Airbnb rentals) variations with respect to the important explanatory variables that significantly influence the room price of peer-to-peer accommodation.
Pricing and revenue management are the two most frequently researched areas in the hospitality domain. Due to the theoretical and practical criticality of room pricing, Airbnb hosts must master room pricing to increase their profitability after satisfying guest expectations. Traditional corporate hotels and hotel chains, as well as peer-to-peer rentals, are easily searchable online. At the same time, in the absence of reputational corporate brand identity in Airbnb shared rentals, the hosts must provide some information for purposes of increasing the attractiveness of their property in terms of amenities and other pricing criteria, which are comparable and searchable. Thus, hosts in Airbnb not only fix their price but also decide the terms and conditions for providing room and services to improve their business and profits.
On the other hand, guests are also at their liberty to select the property that is most economically advantageous and serves their need satisfactorily. Every day, thousands of guests select not to rent a traditional hotel room and instead spend the night in a room listed on peer-to-peer rentals by a stranger. In the literature of hospitality and tourism, studies have shown the increasing trend of novelty seeking and collaborative consumption on sharing rental platforms based on travel bragging items (Guttentag, 2015). Tourists and guests are looking for specific amenities items offered by the hosts on these platforms (Abrate and Viglia, 2017). However, there is a lack of studies on room pricing determinants for online shared rentals. Even traditional hoteliers have paid less attention to important determinants that influence realistic room price decision (Hung et al., 2010). Therefore, the objective of this work is to investigate and identify the key determinants from various offered amenities, accompanied by some specific price indicators identified in the literature that are influential in room pricing. Additionally, this study will explore any potential city-specific generalizations of the key determinants’ influence on room pricing using the Airbnb listings dataset of 11 different US cities. To the best of our knowledge, this is the first endeavor to explore key determinants for generalizability across cities in the same country, based on 143 explanatory variables relating to amenities and few other price indicators.
Also, there is a research gap regarding any generalization of key determinants’ influence on room price across cities in the same country on peer-to-peer rentals. In a related work (Wang and Nicolau, 2017), room pricing determinants in 33 Airbnb cities are shown to be the same as for traditional hotels (Chen and Rothschild, 2010). Another study explored the main pricing factors from a limited dataset comprising 794 samples of Airbnb listings from a single city—metropolitan Nashville, Tennessee—and concluded distance from convention center as a point of sensitivity for room pricing (Zhang et al., 2017). Some of the studies confirmed the positive and significant influences of reviews, ratings, host photos, and Superhost status on room pricing (Liang et al., 2017). Due to the popularity of the Airbnb, both hosts and guests are likely to see sharing rental as cheaper in price and offer a novel experience (like igloos, castles or tree houses) than the traditional hotel accommodation. This experience as an user motivation, stimulates guests “utilitarian value” or hedonic value (Prebensen and Rosengren, 2016; Miao et al., 2014).
The success of a regression-based hedonic pricing model depends on the validity of assumptions associated with these models. However, when the dataset involves nonlinearity or hidden multicollinearity, nonparametric models are more suitable. Therefore, the present study applied RF and CTree as decision tree-based nonparametric data-mining models to deal with the nonlinearity issues in the dataset (Janitza et al., 2016). The present study employed a comparative approach by using three different methods—OLS, random forest (RF), and conditional inference tree(CTree)—applied on a vast quantity of data from the Airbnb listing dataset for 11 cities in the US (151,955 observations and 143 explanatory variables).
While most studies have usually focused on city scale we analysed price determinants across 11 cities in the same country (US). Therefore, this study extended hedonic pricing analysis at the country level comparing the results obtained from 11 different cities.
Therefore, the two main novelty aspects of the present work are: investigating the overall as well as local, city-specific dissimilarity in the relationship between room pricing and its main determinants for Airbnb rentals and application of two machine learning algorithms (RF and CTree) for exploring room price determinants.
The aim of the research was to compare the performance of the applied models that identify the influence of room price determinants, and their estimated results, to contribute to the existing literature on sharing accommodation
The present work primarily contributes to the identification of the important determinants influencing room pricing for Airbnb rentals and to discern any possibility of city-specific generalization for identified determinants in the hospitality and tourism industry. Because the long-term success of the rental industry is critically determined by pricing (Hung et al., 2010; Zervas et al., 2017), the room pricing determinants help Airbnb hosts to develop better practices by recognizing the pricing determinants that persuade guests’ readiness to pay and then fixing the room price based on guests’ insights and fulfillment of expectations. Moreover, this study explores whether the room pricing determinants are similar or locally generalizable across 11 cities in the US, which may help Airbnb hosts to undertake city-specific room pricing strategies and to develop further strategies to improve their services regarding identified main determinants that influence pricing in the underlying dynamic market situations. It is essential for Airbnb hosts to communicate the listing attributes effectively, in terms of amenities and other features that satisfy guests, and set prices for long-term success.
The rest of the study is organized as follows: The next section includes a comprehensive survey of relevant research works. Then we present the methodology, including the dataset description. Thereafter, we applied the OLS, random forest, and decision tree methods to investigate the determinants’ influential roles in affecting room price. The final section presents implications and conclusions.
Section snippets
Literature review
The appearance of ‘the sharing economy’ has resulted in unique growth with respect to a number of users, empowering innovative areas of socioeconomic interaction (Sundararajan, 2016). In the last few decades, the information communication technology revolution and the diffusion of smartphones and social media have tremendously influenced the digital sharing economy (Anderson, 2014; Cohen and Kietzmann, 2014; Zervas et al., 2017).
In literature there are contradictory opinions and assessments on
Methodology
The present research comprises two phases. In the first phase, three methods—traditional OLS, random forest tree, and conditional inference trees —were applied to test their success in modeling the relations between 143 explanatory listing attributes and room price. In the next phase, we investigated the variables’ importance to identify the key determinants having a significant influence on the room price.
In statistical computing, variable importance in regression is a well-researched topic (
Results and discussion
The open-source R environment (R Foundation for Statistical Computing, 2016) on a Windows system with an Intel core and 2.5 GHz processor was used for conducting all experiments applying the OLS regression, random forest, and CTree models. The results of the variable influence of 143 explanatory variables (described in Section 3.2) on room pricing were estimated using OLS regression applied to full samples of the whole dataset, and separately, the 11-individual city-specific individual samples
Theoretical and practical implications
The study contributes to the extant literature in the following ways: it contributes to the tourism & hospitality management domain for contemplating the accommodation industry as small-travel and tourism business (Wang and Hung, 2015). The focus of the study is to identify key price determinants from a huge number of host listing attributes of a sharing rental platform and to explore city-specific generalization for identified determinants across eleven US cities. As we have mentioned both in
Conclusion
The study developed an approach to identify the relationships between determinants and room pricing in Airbnb, peer-to-peer rentals using six variables chosen based on previous literature and 137 amenities offered by the hosts. The work Implemented OLS regression using full samples of the listing dataset and found 53 explanatory variables significantly determining the room pricing. To explore better model fitment, we applied two tree-based models: random forest tree and CTree. Unlike OLS, these
Acknowledgement
The authors would like to thank the editor and the anonymous reviewers for providing valuable feedback through their constructive comments during revising the manuscript.
References (87)
- et al.
Prioritisation of the hotel attributes according to their influence on satisfaction: a comparison of two techniques
Tour. Manag.
(2015) - et al.
Purpose of stay and willingness to stay as dimensions to identify and evaluate hotel experiences
Int. J. Hosp. Manag.
(2013) - et al.
A preliminary study on customer expectations of hotel hospitality: influences of personal and hotel factors
Int. J. Hosp. Manag.
(2012) - et al.
Diagnosing and correcting the effects of multicollinearity: bayesian implications of ridge regression
Tour. Manag.
(2019) - et al.
Dynamic room pricing model for hotel revenue management systems
Egypt. Inform. J.
(2011) - et al.
Being better vs. Being different: differentiation, competition, and pricing strategies in the Spanish hotel industry
Tour. Manag.
(2013) Why are flexible booking policies priced negatively?
Tour. Manag.
(2018)- et al.
Trust and reputation in the sharing economy: the role of personal photos in Airbnb
Tour. Manag.
(2016) - et al.
Variable selection using random forests
Pattern Recognit. Lett.
(2010) What makes an Airbnb host a superhost? Empirical evidence from San Francisco and the Bay Area
Tour. Manag.
(2018)