How Do Reviews Impact Airbnb’s Prices? A Hedonic Approach

Abstract

The travel accommodation sector within the sharing economy relies heavily on user-generated reviews. Drawing on data from insideairbnb.com for the Porto district from 2016 to 2020, this study examines the influence of online reviews from the standpoint of the sentiment expressed on accommodation prices, alongside other determinants such as locational attributes. The primary objective is to assess a broad set of factors affecting listing prices, with a particular focus on the degree and nature of sentiment expressed in online reviews. The dataset, comprising more than 250,000 reviews, was enriched with spatial and geographical variables, including key amenities, accessibility to public services, host characteristics, and locational indicators. A hedonic spatial regression model was employed to account for spatial dependencies. The findings reveal that sentiments expressed in user reviews exert a stronger influence on pricing than purely quantitative review metrics. Furthermore, host and listing characteristics, as well as geographical factors, play a substantial role in determining prices. The main contribution and novelty of this study lies in the joint analysis of sentiment and geographical attributes as drivers of accommodation pricing. Another contribution of this paper lies in the analysis of a broad geographical area encompassing both a historic city that is popular among European destinations and predominantly rural regions.

1. Introduction

The Airbnb sector has transformed the hospitality industry beyond recognition, driven by pivotal and disruptive changes (S. Lee & Kim, 2023; Tafesse & Tariq, 2025), including low-cost pricing strategies and highly personalized and flexible interactions between accommodation hosts and guests (Sainaghi & Chica-Olmo, 2022; Suess et al., 2020; Abrate & Viglia, 2022). Abrate and Viglia (2022) categorize the Airbnb sector as the “biggest player in the lodging industry,” while Voltes-Dorta and Sánchez-Medina (2020) describe it as the leading challenger to both traditional hotels and other accommodation providers.

The emergence of new forms of shared accommodation, along with the increase in hostels and B&B establishments, has enabled a growing number of relatively less affluent consumers to travel for leisure several times per year, either domestically or internationally (Chica-Olmo et al., 2020; Zervas et al., 2021). Demonstrating resilience during the COVID-19 crisis (Más-Ferrando et al., 2024), Airbnb benefitted from lower prices and greater compliance with social distancing measures, proving its value to risk-averse tourists and contributing to the sector’s survival in highly challenging circumstances.

The Airbnb sector offers a wide range of options in terms of price, features, and locations, whether in urban areas near cultural centers and lively neighborhoods or in proximity to natural parks and low-density territories. Airbnb aligns with contemporary shifts in tourists’ preferences toward greater flexibility in location and services, as well as autonomy and personal freedom in their travel experiences, all at lower costs (Huang & Huang, 2024). The platform has created a new type of tourism demand by delivering “value” and expanded choices to budget-conscious tourists willing to forgo certain extras and conveniences for more affordable accommodations.

Price, a key criterion in choosing hotels, has long been a focus of research in the accommodation industry, including the Airbnb sector (D. Wang & Nicolau, 2017; Ert & Fleischer, 2019; Voltes-Dorta & Sánchez-Medina, 2020). It is widely acknowledged that a range of accommodation attributes, location-specific factors, and host characteristics significantly influence prices across the industry and particularly within the Airbnb sector (Perez-Sanchez et al., 2018; Lawani et al., 2019; Gyódi & Nawaro, 2021; Gibbs et al., 2018). Online reviews, including guests’ ratings published on travel websites, have likewise been identified as important determinants of pricing (Öğüt & Onur Taş, 2012; Blal & Sturman, 2014).

Although previous research has successfully identified several causal relationships (Gibbs et al., 2018; Lawani et al., 2019) between price variations and Airbnb-specific characteristics—such as the number of reviews and quantitative ratings reflecting tourists’ experiences—far less attention has been given to simultaneous analyses of technical attributes and the emotional aspects reflected in the sentiments expressed in tourists’ online reviews.

Several gaps and under-researched topics remain in the literature. Y. Chen and Xie (2017), Gibbs et al. (2018), Sainaghi et al. (2021), and Leoni and Nilsson (2021) emphasize that accurate pricing is crucial for the Airbnb platform’s profitability, particularly given the sector’s low prices and often low occupancy rates. Moreover, the ongoing expansion of the sector (Leoni & Nilsson, 2021; Abrate & Viglia, 2022), driven by the conversion of existing housing stock into sophisticated tourist accommodations or the development of new facilities (Gyódi & Nawaro, 2021), has resulted in significant investments and high fixed costs, demanding a more professional approach to pricing and greater information availability. Despite the high degree of heterogeneity within the sector, Airbnb is becoming increasingly professionalized, requiring technically robust analyses of price determinants on the part of hosts.

A large number of papers discuss the short-term rental price strategies. See in this regard Pérez-Rodríguez et al. (2024), Zheng et al. (2023), and Di Persio and Lalmi (2024). However, the current literature on short-term rental pricing strategies is considerably absent in the following aspects. The literature on Airbnb listing prices often focuses on clearly delimited urban settings that share a common ambiance, such as a homogeneous urban landscape containing well-identified cultural facilities and major attractions. “While a substantial amount of research is available to identify the factors that impact Airbnb’s listing prices in several cities around the world, the case of the district of Porto in Portugal remains relatively unexplored (Milunovich & Nasrabadi, 2025). Considering the recent exponential growth of the Airbnb segment in the European context, with new firms entering the market at an accelerated pace, and the ongoing debate about the negative impacts of the sector on the housing market and on the profitability of traditional hospitality sector, understanding Airbnb’s price drivers is critically important to assess the likelihood of further damage to both the housing market and the traditional accommodation sector.

Moreover, the recent expansion of Airbnb through the inclusion of both individual and professional hosts has intensified internal competition within the sector. This heightened competition favors the most cost-effective units (Dogru et al., 2020), suggesting that further research into price determinants is necessary to help property owners invest strategically in the most in-demand amenities.

While numerous studies have explored price determinants, this paper goes further by analyzing not only the influence of physical attributes, guest ratings, host characteristics, and location effects but also the relatively under-researched issue of how sentiment expressed in reviews affects price and how spatial factors drive similarities in pricing strategies among geographically related listings. Overall, this study aims to evaluate the influence of various listing attributes on pricing, including the sentiments expressed in reviews. Additionally, it investigates price determinants across a large geographical area that encompasses a major tourist city, its surrounding suburbs, and second- and third-tier cities in rural areas—a research context that remains relatively uncommon. The main purpose of this research is therefore to analyze a broad range of factors influencing listing prices, with a particular emphasis on the degree and nature of sentiments expressed in online reviews, as well as on a variety of location indicators.

This paper is organized as follows: Section 2 offers an overview of the relevant literature; and Section 3 presents the methodology used. Empirical models and results are presented in Section 4. Finally, Section 5 unveils the conclusion, limitations, and future work.

2. Literature Review

2.1. Airbnb Price Determinants

The literature on the Airbnb phenomenon has extensively investigated the factors influencing the pricing of Airbnb listings (Lin & Yang, 2023; Y. Yang et al., 2022). This body of research has consistently demonstrated that a wide range of variables—including location-specific attributes, host characteristics, listing types, and review ratings—exert significant effects on Airbnb prices (Teubner et al., 2017; Shokoohyar et al., 2020; Sainaghi et al., 2021). However, review scores and the number of reviews serve as important quality signals, providing insights into the host’s reliability and performance, thereby reducing information asymmetry in the marketplace (Boto-García, 2022). Falk et al. (2019) note that previous studies have concluded that the prices of Airbnb accommodation are determined by a combination of functional characteristics of the listings (Y. Chen & Xie, 2017; Gibbs et al., 2018; D. Wang & Nicolau, 2017), host attributes and reputation (Ert et al., 2016; Y. Chen & Xie, 2017; Teubner et al., 2017; Magno et al., 2018), booking and cancellation policies, personal reputation (Mauri et al., 2018), and the availability of local amenities (Gunter & Önder, 2017).

Five key determinants of Airbnb listing prices are frequently examined in the literature: listing attributes, host attributes, listing reputation, rental policies, and location (Cai et al., 2019). Listing attributes have been shown to significantly influence Airbnb prices, although the magnitude and complexity of these effects vary considerably. Factors such as the size of the property, the number of bathrooms, and the offering of an “entire place” generate utilitarian value for Airbnb guests, thereby contributing to higher prices (Lladós-Masllorens et al., 2020). Similarly, amenities such as air conditioning, free internet, and free parking have a measurable impact on listing prices (Dudás et al., 2020). In the higher-end market segment, features like included breakfast, access to outdoor spaces designated for smokers, and availability of high-speed Wi-Fi tend to increase prices, whereas in the lower-end segment, the number of bedrooms and bathrooms, as well as free parking, play a more significant role in price determination (Yoong et al., 2025).

Host attributes, including years of hosting experience, designation as a Superhost—a signal of quality (Gyódi & Nawaro, 2021)—the number of listings managed, reputational effects, and verified identities, are also associated with higher prices (Magno et al., 2018; D. Wang & Nicolau, 2017; Teubner et al., 2017). Another crucial factor influencing price formation is the strictness of rental policies. Guests often perceive stricter policies as indicators of higher accommodation quality and greater host involvement in providing services and addressing potential issues (Lladós-Masllorens et al., 2020; Gibbs et al., 2018; Luo & Tang, 2019).

Research indicates that Airbnb guests are generally younger, technologically savvy, and less reliant on traditional sources of information (Dogru et al., 2020; Possamai, 2022). Consequently, repurchase intentions among Airbnb users are strongly influenced, both directly and indirectly, by electronic word-of-mouth (eWOM) (Mao & Lyu, 2017), the quality and attractiveness of host websites, and the content and positivity of online reviews (Forgacs & Dimanche, 2016). Airbnb’s success is also rooted in a continuous stream of technological innovations, such as the creation of the “Superhost” status to identify top-performing hosts, and in efforts to attract business travelers and a diverse range of customer segments (Guttentag et al., 2018; Gibbs et al., 2018).

Online reviews have evolved to play a pivotal role in the Airbnb ecosystem. When analyzing reviews, customers tend to evaluate lodging units with stronger reputations and higher online visibility more favorably (Hu et al., 2008). Review content is critical in shaping purchasing decisions. During the initial stages of decision-making, tourists typically examine listing prices as well as property features and amenities. However, initial assumptions and impressions based on price and advertised features are often confirmed or contradicted by the sentiments expressed in other users’ reviews. Therefore, understanding the impact of eWOM is crucial for comprehending consumer decision-making and behavior, and, by extension, for developing effective pricing strategies (Chevalier & Mayzlin, 2006; Lin & Yang, 2023). As Lin and Yang (2023, p. 1) note, “reviews can be helpful to mitigate asymmetric information (Park & Nicolau, 2015; Manes & Tchetchik, 2018) and enhance a product’s reputation (Jalilvand et al., 2017).”

While earlier studies have extensively analyzed review scores and the number of reviews, more recent research emphasizes the importance of review content in influencing consumer purchasing decisions (Chevalier & Mayzlin, 2006; Li & Hitt, 2010; Nowak & Smith, 2017; M. Cheng & Jin, 2018; Lawani et al., 2019; Lin & Yang, 2023). The existing literature suggests that most reviews focus on three principal topics: location (often measured as the distance to the city center, typically the most influential driver of price), amenities, and host-related attributes (von Hoffen et al., 2018; Tussyadiah & Zach, 2017; M. Cheng & Jin, 2018; Zhu et al., 2019; Cavique et al., 2022; Lin & Yang, 2023; Guttentag et al., 2018). Consumers increasingly rely on the substantive content and sentiment of online reviews rather than solely on numerical ratings or scores.

Jiang et al. (2024) found that the impact of negative reviews on prices is twice as substantial as that of positive reviews. Notably, the effect of negative reviews is even more pronounced for lower-priced listings, where negative sentiments exert a fourfold influence on price compared to positive reviews. Regarding review topics, Lin and Yang (2023) concluded that reviews focused on listing characteristics and location exert the most significant influence on prices. This finding suggests that customers place considerable importance on room-specific features and the neighborhood context of the property.

Voltes-Dorta and Sánchez-Medina (2020) categorize the explanatory variables affecting prices in traditional hotel establishments into three groups: (1) hotel features and amenities, such as room size, Wi-Fi, television, or a gym (C. F. Chen & Rothschild, 2010; Schamel, 2012); (2) quality signals, including the hotel’s brand, star rating, and customer reviews (Masiero et al., 2015; Schamel, 2012; Y. Yang et al., 2022) and (3) locational factors, such as proximity to the city center and other tourism hotspots. While proximity is generally expected to increase hotel prices (Soler & Gemar, 2018), it is also possible to find lower-priced hotels in downtown areas due to intensified spatial competition among providers (C. F. Chen & Rothschild, 2010).

Attributes frequently highlighted in guest reviews are related to location (proximity to points of tourist interest and neighborhood characteristics), host attributes (service and hospitality), and property characteristics (facilities and atmosphere) (Tussyadiah & Zach, 2017; X. Cheng et al., 2019). Guests’ assessments of “authenticity” and “home benefits,” as defined by So et al. (2018) as “functional attributes of a home—household amenities, homely feel, and large space similar to those coming from a home environment,” and by Guttentag (2016) as a “feeling of being at home while staying in a hotel and access to practical residential amenities such as a full kitchen, a washing machine, and a dryer,” enhance the perceived sense of homeliness and continuity with guests’ everyday lives.

Guest comments are not limited to evaluations of the accuracy of the listing information, the cleanliness and homeliness of the accommodations, host attributes, or overall satisfaction with their stay. Reviews often include practical information for prospective guests, such as the availability of maps and geographical data, tips and recommendations for local shopping to facilitate access to essential supplies upon arrival, and assessments of hosts’ personal touches that enrich the overall experience (Zamani et al., 2019).

In relation to locational factors, the literature lacks studies on the determinants of prices in geographical settings that combine major urban areas and their neighboring rural hinterlands, both of which are dependent on the same major airport as an entry point, but different in many other respects (Falk et al., 2019; Voltes-Dorta & Sánchez-Medina, 2020). Mixed urban–rural areas are notoriously absent from the literature. The hospitality market, in general, and Airbnb market, in particular, is characterized by strong spatial dependence (Gyódi & Nawaro, 2021), as listings in a given delimited area share a set of common attributes (e.g., access to specific tourist attractions, ambience of the area, etc.) that similarly affect price (Zhang et al., 2017; Tang et al., 2024). Therefore, in this study, we consider spatial effects by examining a large geographical setting that combines urban and rural areas.

The examination of price determinants is important for additional reasons. Sainaghi et al. (2021) describe most Airbnb hosts as “amateur innkeepers,” with limited pricing expertise, particularly regarding dynamic pricing and revenue management (Y. Chen & Xie, 2017). Hosts’ pricing strategies are sometimes irrational (Xie & Kwok, 2017), which is unsurprising given that many Airbnb hosts are micro-entrepreneurs with little prior experience in the hospitality sector and are often content with modest increases in income. One of the key factors behind Airbnb’s recent expansion is the opportunity it offers homeowners to earn additional income (Lin & Yang, 2023). Understanding the features that most positively impact prices provides valuable insights into future profitability, investment priorities, and strategies to maximize returns. Consequently, “the role of pricing and the effect of such strategies on agents’ performance deserve substantially more attention” (Xie & Kwok, 2017; Leoni & Nilsson, 2021, p. 2).

Given that Airbnb’s economic appeal continues to rely on low prices, and price is recognized as the core competitive advantage of peer-to-peer (P2P) offerings (Tussyadiah & Pesonen, 2016; Gibbs et al., 2018), being frequently cited as the primary reason for choosing Airbnb over traditional hotels (M. Kim et al., 2025; Tussyadiah & Pesonen, 2016), it is essential to understand which investments should be prioritized to achieve an optimal balance between tourists’ expectations regarding price and amenities and hosts’ expectations for profitability, which is, in turn, dependent on the cost of the amenities and services offered.

2.2. Sentiment Analysis

The Airbnb phenomenon has fundamentally challenged long-established hotel business models (Gyódi & Nawaro, 2021), particularly in terms of the influence of tourists’ feedback on hotels’ efforts to improve the quality of the servicescape and responsiveness to customer needs through efficient complaint resolution mechanisms and enhanced service quality (Teubner et al., 2017; Pérez-Rodríguez et al., 2024; Nieto García et al., 2020; Y. Yang et al., 2022; Gyódi & Nawaro, 2021).

Numerous studies have focused on Airbnb’s attributes and characteristics. The literature reveals a general consensus regarding the main dimensions of service quality within the Airbnb context, namely, communication, experience, location, product/service, and value (X. Cheng et al., 2019; Guttentag et al., 2018). Although research indicates that users often evaluate Airbnb based on practical attributes such as location—sometimes at the expense of experiential attributes like social interaction (Guttentag et al., 2018; Tussyadiah, 2016)—memorable Airbnb experiences are frequently rooted in social interactions, warm welcomes, and the helpfulness demonstrated by hosts (Sthapit & Jiménez-Barreto, 2018; Priporas et al., 2017). Unlike interactions with hotel staff, guests often highlight and praise hosts’ personal qualities and social attributes (X. Cheng et al., 2019). In many cases, these positive perceptions are explicitly expressed in online reviews.

Research shows that online reviews significantly influence hospitality users’ choices and behaviors (Matzat & Snijders, 2012), affecting decisions regarding both price and specific attributes. Liang et al. (2017) found that electronic word-of-mouth (eWOM) plays a vital role in shaping consumers’ perceived value, perceived risk, and repurchase intentions concerning accommodation services. Online reviews enhance the quality of interactions and information exchange between hosts and travelers on the Airbnb platform (Zhou et al., 2018). Reviews reflect past customer experiences, often serving as the only source of information available to infer the price–quality ratio for new products and services (Weismayer & Pezenka, 2019). An area receiving increasing scholarly attention in the Airbnb sector is the content and sentiment of online reviews. Sentiment analysis is defined as the process of using natural language processing (NLP), text analysis, and computational linguistics to identify and extract subjective information from textual data, such as opinions, emotions, or attitudes. Its primary aim is to determine the polarity of sentiment within a piece of text—typically categorized as positive, negative, or neutral. In this regard, we follow Puh and Babac (2023) and Awotunde et al. (2023), guided by considerations of accuracy in identifying the predominant sentiment. More advanced sentiment analysis tools can detect specific emotions (e.g., joy, anger, sadness), measure the intensity of sentiment (e.g., very positive, slightly negative), and identify the target of the sentiment (e.g., a particular product or person). Sentiment analysis is widely employed in contexts such as product review analysis, brand reputation monitoring on social media, customer feedback evaluation, and market research. In its simplest form, sentiment analysis is framed as a two- or three-class problem, classifying text as positive, negative, or possibly neutral.

Patel and Arasanipalai (2021) note that the exponential growth in social media content has driven the development of sophisticated methods for automatically identifying customer sentiments. Such methods aim to analyze tweets, posts, and comments not only for overall sentiment polarity but also for specific emotional content (e.g., anger, sadness, happiness).

The quality and assessment of the tourist experience are critical for tourism services, as higher satisfaction levels derived from positive experiences lead to stronger intentions to return and more favorable word-of-mouth promotion. By analyzing travelers’ emotions and experiences, as expressed in online reviews through data mining and text analysis techniques such as sentiment analysis, researchers have made significant theoretical contributions to understanding the tourism experience and its implications for repeat visitation (Moro et al., 2019).

Alsudais and Teubner (2019) argue that sentiment analysis is an effective technique for classifying the emotional content of online reviews, sentences, complaints, and messages. When examining a dataset of 2,686,354 reviews and 12,353,382 sentences from Airbnb listings in 15 U.S. cities, they found that 98.1% of reviews and 76.4% of sentences were positive. In contrast, only 1.06% of reviews and 4.7% of sentences expressed negative sentiments.

Martinez et al. (2017) approached sentiment analysis from a different perspective, aiming to identify which review features, including sentiment and review length, had the most significant impact on occupancy rates. Their findings revealed that while longer reviews positively influenced occupancy figures, the positivity of the review summaries themselves had no significant effect. Instead, factors such as the number of reviews, the number of listed amenities, and the length of the online review summary were found to be stronger indicators of occupancy rates.

2.3. Hedonic Price Modeling

Most studies examining the pricing of Airbnb listings are based on the hedonic price model (Gyódi & Nawaro, 2021; Más-Ferrando et al., 2024), one of the most widely used methods in this field of research. For instance, Santos (2016) conducted a hedonic price analysis based on a global dataset of 8000 hostels, demonstrating that accommodation prices were largely explained by hostel characteristics as rated by guests. This study’s main findings indicated that cleanliness, location, and amenities were the most influential factors in determining prices. Similarly, Gibbs et al. (2018), in their analysis of Airbnb pricing records from five Canadian cities, reported significant impacts of physical features, location, and host attributes on prices. Their findings indicated that guest privacy is highly valued, while a greater number of reviews often correlates with lower prices, possibly reflecting increased demand for affordable options. Locational factors, measured by proximity to downtown areas, also exerted a notable effect, with prices decreasing as distance from the city center increased. Moreover, Superhost status was associated with price premiums in specific geographic contexts, whereas independent properties were often compelled to set lower prices compared to multi-unit dwellings.

Hedonic price modeling is an econometric technique used to estimate the value of a good or service by decomposing it into the values of its constituent attributes. Sirmans et al. (2005) note that hedonic models, typically estimated via single-stage regression equations, analyze how the various characteristics of a good or service influence its price. According to Gyódi and Nawaro (2021) and Más-Ferrando et al. (2024), the hedonic pricing model, grounded in Lancaster’s characteristics theory, explicitly assumes that a product’s price reflects consumers’ valuation of its diverse attributes. The approach presupposes that the price of a complex product, such as a holiday rental property, is effectively the aggregate of monetary valuations assigned by consumers to each of its key attributes, such as the number of bedrooms, location, and square footage. Thus, the total price represents the combined value of these measurable attributes, as posited in Lancaster’s seminal work (Lancaster, 1966). Lancaster’s Characteristics Theory of Value asserts that products and services derive value from the benefits conferred by their attributes, rather than from the products as indivisible wholes. Researchers commonly employ regression analyses to quantify each attribute’s contribution to the overall price.

The earliest applications of hedonic price analysis to the Airbnb sector explored the effects of variables such as property size, guest rating scores, listing popularity, host characteristics, amenities, and distance to the city center (Gibbs et al., 2018; D. Wang & Nicolau, 2017). These studies confirmed that additional space (measured in square meters), favorable guest reviews, and proximity to city centers all contributed to higher prices and price premiums. However, the impact of the number of reviews on listing prices has proven more nuanced. In several instances, a higher number of reviews was associated with a statistically significant decrease in price, suggesting possible reverse causality (Gyódi & Nawaro, 2021).

Sainaghi et al. (2021) emphasize that analyzing price determinants in the Airbnb market is inherently complex, given the overlapping and interactive effects of multiple variables, including listing features, host attributes, and locational factors. Prior studies on Airbnb have consistently demonstrated that listing characteristics (such as size and number of rooms), locational attributes (including distance from the city center, proximity to cultural attractions, and accessibility via public transportation) (Gibbs et al., 2018), overall quality, and host reputation are statistically significant predictors of price. These findings suggest that the price impact of individual attributes can be understood in an additive manner.

Gyódi and Nawaro (2021) notably found that greater distance from the city center or metro stations often corresponded to higher listing prices, contradicting earlier assumptions that proximity would uniformly increase prices. Reviews by Gyódi and Nawaro (2021), D. Wang and Nicolau (2017), and M. Kim et al. (2025) offer comprehensive overviews of hedonic price studies, concluding that hotel and Airbnb accommodation prices are primarily determined by location factors, quality as measured by class ratings, property size, customer reviews, amenities, and other property characteristics.

Within these studies, researchers have tested the impact of location variables such as distance to the city center (Bull, 1994), proximity to metro stations or central railway stations (Thrane, 2007), and adjacency to key tourist districts (Carvell & Herrin, 1990). S. K. Lee and Jang (2011) further analyzed the importance of accessibility to airports and central business districts. Deboosere et al. (2019) demonstrated that access to public transportation and neighborhood socioeconomic factors significantly influenced Airbnb prices in New York City. Similarly, Y. Yang et al. (2022) showed that accessibility—as measured by distance to metro stations, airports, and universities—affected guest satisfaction levels. Perez-Sanchez et al. (2018) found that listings located in scenic sightseeing areas or near coastlines commanded premium prices in Spanish cities along the Mediterranean.

Not surprisingly, location or neighborhood characteristics remain among the most extensively analyzed dimensions in both hedonic price models and in the decision-making processes of potential Airbnb guests searching for suitable accommodations. Lin and Yang (2023) observed that listings located closer to city centers were less susceptible to the negative impacts of unfavorable reviews compared to those situated in more remote areas. Studies by Gibbs et al. (2018) and Lawani et al. (2019) further underscore the significance of location in shaping overall guest experience evaluations.

According to Gibbs et al. (2018) and Lorde et al. (2018), physical features of the property and host characteristics have considerable influence on prices, while an increased number of reviews is often correlated with lower prices.

Hedonic pricing is a well-established method for analyzing consumer preferences based on the attributes influencing product prices, particularly within urban housing markets and accommodation choices (e.g., Helbich et al., 2014). The literature on Airbnb pricing has extensively explored the factors driving price formation in the context of short-term rental listings (Hung et al., 2010; S. K. Lee & Jang, 2011; S. Yang & Ahn, 2016). Previous studies have consistently corroborated the influence of location, host characteristics, listing type, and review ratings on pricing (Teubner et al., 2017; Shokoohyar et al., 2020; Sainaghi et al., 2021).

In the specific context of Airbnb, researchers including S. Yang and Ahn (2016), Y. Chen and Xie (2017), So et al. (2018), Fearne (2022), and Ghosh et al. (2023) have adopted spatial hedonic price modeling to better understand consumer choices and the valuation of different listing attributes. Zhang et al. (2024) and Y. Wang et al. (2020) have demonstrated the high accuracy and effectiveness of hedonic pricing methods in estimating consumer preferences within the Airbnb market (Östh et al., 2025).

Moreover, Lawani et al. (2019) applied a spatial autoregressive hedonic model to Airbnb listings in Boston, integrating sentiment analysis to assess quality scores derived from user reviews. Their results indicate that prices are influenced by review scores, room characteristics, and neighborhood attributes.

Based on the literature review and comments integrated into this section, we formulate the following hypothesis:

H1: 

Higher levels of sentiment expressed on online reviews are associated with higher prices;

H2: 

A more central location increases the willingness to pay;

H3: 

Attributes such as access to a swimming pool or public transport tend to command higher prices;

H4: 

The level of expertise and professionalism, reflected in the number of years of experience, is associated with higher pricing.

3. Methodology and Model Specification

3.1. Methodological Strategy

Regarding the impact of proximity and spatial effects in the context of hedonic pricing, it is important to note that spatial spillover effects are typically not accounted for in classic models. A standard approach for addressing spatial autocorrelation in hospitality research is the geographically weighted regression (GWR) method. GWR assumes that relationships between variables vary across locations and generates localized regression coefficients for each observation (M. Kim et al., 2018). GWR has proven valuable in analyzing local variations in estimated coefficients and has been employed in hotel price studies to highlight the significant influence of distance to transportation infrastructure (M. Kim et al., 2018; Soler & Gemar, 2018; Zheng et al., 2023) and proximity to tourist attractions (M. Kim et al., 2018; Zheng et al., 2023). For example, Zhang et al. (2017) applied GWR to analyze Airbnb prices in Nashville, using a limited set of variables, including distance from a convention center and highways.

In this study, we employ a spatial model to control for spatial effects. Specifically, we use the spatial autoregressive (SAR) model, which accounts for substantive spatial dependence (Anselin, 1988). The SAR model is specified as follows:

$$\mathrm{P}=\rho WP+Q\beta +e$$

(1)

where P represents the listing price in its logarithmic form. Following previous studies in this area (Thrane, 2007), we employ a log-linear specification for the pricing function rather than a linear one. The symbol ρ represents the coefficient of the spatial component; W refers to the spatial distance matrix using 1 km as the distance threshold; Q = (k = 1, …, m) represents each of the specific attributes of the listing; β stands for the vector of the coefficients of the explanatory variables; and e is the independent error term following a normal distribution.

As the listing prices are predicted to be spatially dependent, as proprieties cluster together, we defined a spatial weighting matrix that puts the same positive weight on continuous locations and a zero weight on all other counties. The spatial information on administrative boundaries, like counties, was downloaded as a shapefile (.shp) from an official site. For further information, please see the entry spregress in the STATA manual.

Past research has emphasized the significance of numerous specific attributes in determining listing prices, including cleanliness (Thomsen & Jeong, 2021), guests feeling welcome (Tussyadiah & Zach, 2017), proximity to key attractions (Kiatkawsin et al., 2020), host responsiveness (Zhang et al., 2017), and clarity of information provided on the website (C. K. H. Lee et al., 2023). Moreover, several studies have distinguished between positive and negative reviews to identify experiential aspects (e.g., noise) that generate either favorable or unfavorable Airbnb experiences (M. Cheng & Jin, 2018; Ju et al., 2019; Luo & Tang, 2019; Zhang et al., 2017). Some researchers have adopted specific conceptual frameworks to analyze Airbnb guest reviews—for instance, Johnson and Neuhofer (2017) examined Airbnb experiences through the lens of value co-creation, while Bao et al. (2022) explored such experiences using the Experience Economy Model. In our research, we adopt the methodological approach outlined by Guttentag et al. (2018). Notably, in this study, the variable sentiment is used as a proxy for Airbnb guests’ perceived satisfaction with the accommodations they select.

This analysis is based on Airbnb data for the Porto Metropolitan Area, encompassing 17 municipalities. The Porto District, located in the northwest of Portugal at the mouth of the Douro River, covers an area of approximately 2455 km² and has a population of around 1,855,940 inhabitants as of 2023. As a prominent tourist destination in the European context, the region—or at least the broader North zone—welcomed 6.3 million international tourists in 2024. The city of Porto itself is a magnet for urban, cultural, and architectural tourism, hosting several UNESCO-recognized landmarks such as the Historic Centre, the Dom Luís I Bridge, the Palácio da Bolsa, Livraria Lello, and the Cathedral. Porto is served by Francisco Sá Carneiro Airport, located approximately 11 km northwest of the city center, and by an efficient metro network. The airport, serviced by low-cost carriers such as easyJet and Ryanair, along with its extensive network of direct flights, has helped solidify its role as a vital gateway for tourism flows into the city and the wider region. However, despite the limited number of studies, the importance and relevance of the city are evident. Porto is among the best cities in the world, the city of Porto was distinguished as the Best Urban Destination in 2022, having achieved the eighth place in a list of 50 destinations published by Time Out magazine, according to the AICEP (Portuguese Trade & Investment Agency) (AICEP, 2025).

Data were obtained from the Inside Airbnb website (insideairbnb.com), an independent, non-commercial platform providing tools to explore how Airbnb operates in cities worldwide (Inside Airbnb, 2020). From this source, we retrieved 450,570 reviews spanning the period from 2016 to 2020, along with detailed information on 9588 listings in Porto. The dataset integrates Airbnb data—including listing characteristics, geographic coordinates, and nightly prices—with average sentiment scores extracted from reviews for each listing, based on the sentiment analysis methodology described previously. Additionally, the dataset incorporates GIS-derived variables measuring the distance from each listing to the nearest point of interest, metro station, and coastal location. These geospatial measurements were computed using ArcGIS software version 10.8.1, leveraging multiple shapefiles representing Porto’s geographic features.

A language detection function method of Google Sheets was used to detect the language of the reviews. Of a total of 450,570 reviews, 251,001 were written in English. The dataset was cleansed by eliminating all non-English reviews, reviews with less than 20 characters, automated messages, and reviews containing terms not recognized by the dictionaries used in R.

Table 1. Stages of data cleansing.

Stage	Number of Reviews
Initial reviews	450,570
Non-English reviews	199,569
Short reviews (under 20 characters)	15,957
Automated messages	1533
Terms not recognized	21,333
Cleansed reviews	212,178

details the stages of data cleansing. To ensure linguistic consistency for sentiment analysis, we applied Google Sheets’ language detection function to identify the language of each review. Out of 450,570 total reviews, 251,001 were written in English. We cleansed the dataset by excluding non-English reviews, reviews shorter than 20 characters, automated messages, and entries containing words not recognized by the sentiment dictionaries employed in R. Table 1 summarizes the stages of data cleansing.

In this study, we employed the tidytext package in R, a tool designed for text mining and natural language processing within the principles of tidy data. The package’s primary purpose is to simplify the manipulation, analysis, and visualization of text data by converting it into a tidy format (i.e., a data frame where each row represents a single token—typically a word—and each column represents a variable). This approach facilitates seamless integration with other tidyverse tools and existing text mining packages. For further details, see Silge and Robinson (2017).

Various methods and lexicons are available to evaluate the emotional content of text, many of which are included in the tidytext package. In this paper, we utilized the AFINN sentiment lexicon developed by Nielsen (2011). Unlike most sentiment lexicons that classify words into binary categories, AFINN assigns an integer valence score between −5 and +5 to each term, with negative scores indicating negative sentiment and positive scores indicating positive sentiment. For instance, the word “amazing” receives a score of +4, “good” a score of +3, “unpleasant” a score of −2, and “bad” a score of −3. Nielsen (2011) developed this lexicon using manually labeled posts from Twitter to assess sentiment.

In our study, we applied a quartile analysis to divide the sentiment scores obtained into three categories: low, medium, and high.

3.2. Variables

As noted above, and consistent with Lin and Yang (2023) and prior research, this study controls for a broad range of established price determinants, including review scores, host attributes, listing characteristics, rental rules, and amenities. These variables reflect the principal factors typically included in prior hedonic pricing studies. Furthermore, we calculate linear distances from each listing to three key geographic features.

The independent variables in our analysis are categorized as follows:

• Host Attributes—This includes variables such as the host’s years of experience, the number of listings managed, gender, the language used in the listing description, the acceptance rate, the response rate, and the average response time to guest inquiries. We also consider whether the host holds Superhost status. To qualify as an Airbnb Superhost, a host must host at least ten stays or 100 nights across at least three reservations, maintain a response rate of 90% or higher, keep cancellations under 1%, and achieve an overall rating of 4.8 or above.
• Listing Characteristics—This encompasses the maximum number of guests the listing accommodates, whether the property is an entire place, the number of bedrooms and bathrooms, and amenities such as air conditioning and swimming pool access.
• Location and Nearby Amenities—This group includes variables indicating whether the listing is situated within the municipality of Porto (which hosts the majority of listings) and whether the listing provides precise geolocation data on the Airbnb platform. Additionally, we calculate distances from each listing to the nearest metro station, cultural point of interest, and coastline point.
• Booking-Related Variables—This category covers factors such as the minimum number of nights required, the type of cancellation policy, instant booking availability, the base number of guests included in the price, acceptance of additional guests, and the presence of security deposits or cleaning fees.
• Crowdsourcing Metrics—This set comprises variables such as the average number of customer reviews per month, the overall quantitative rating given by guests, and the sentiment score extracted from textual reviews.

The dependent variable in our analysis is the nightly price of the listing, expressed in logarithmic form, consistent with prior research methodologies. Most data were sourced from insideairbnb.com, except for geospatial variables calculated using ArcGIS Pro 3.1 software and sentiment analysis conducted using R. Spatial regressions were estimated using STATA 17.

Table 2. Descriptive statistics for attribute variables.

	Variable	Description	Obs.	Mean	Std. Dev.	Min.	Max.
Dep. Var.	Price	Log of the price	34,449	1.77293	0.24974	0.82	3.31
Host	Lang	Listing description language (1—English)	34,449	0.6835	0.46512	0	1
	Gender	1—Female; 0—Male	23,424	0.47917	0.49958	0	1
	Host years	Number of years on Airbnb	34,449	3.91622	2.06571	0.227	9.86
	Resp time	Response time (1—under 1 h)	30,010	0.85678	0.3503	0	1
	Resp rate	% of response to users’ queries	30,010	0.96893	0.12431	0	1
	Acceptance rate	% of users accepted by the host	33,244	0.94382	0.16209	0	1
	Superhost	The host has a Superhost status (yes = 1)	34,449	0.37522	0.48419	0	1
	List count	Number of host listings	34,449	13.95277	52.19372	0	550
Booking	Cancelation	Cancellation policy (1—Flexible; 0—Strict)	34,449	0.6408	0.47977	0	1
	Min nights	Minimum nights	34,449	2.29946	5.93665	1	365
	Is location exact	yes = 1	34,449	0.74429	0.43627	0	1
	Security deposit	$	34,228	52.09963	102.7795	0	900
	Cleaning fee	$	34,449	17.84917	18.20115	0	380
	Guests included	Guests included in the listing’s price	34,449	2.06735	1.44453	1	16
	Extra people	$ per extra person	34,449	8.58629	10.02018	0	250
	Instant bookable	yes = 1	34,449	0.78983	1.2789	0	1
	Quarter	Quarters of the year	34,449	2.5992	1.10649	1	4
Location and amenities in the area	Porto	In Porto municipality? (yes = 1)	34,449	0.74324	0.43685	0	1
	Neighbor	Neighbor of Porto municipality? (yes = 1)	34,449	0.17487	0.37986	0	1
	Dist station	Log of the distance (mt) to the nearest tube station	34,449	2.76539	0.54076	0.3597	4.718
	Dist poi	Log of the distance (mt) to the nearest point of interest	34,449	2.70313	0.46624	0.2944	4.027
	Dist coastline	Log of the distance (mt) to the coastline	34,449	3.17842	0.39291	−1.110	4.620
Listing characteristics	Accommodates	Guests the listing accommodates	34,449	4.05446	2.18513	1	17
	Room type	Entire place (yes = 1)	34,449	0.81761	0.38617	0	1
	Nbedrooms	Number of bedrooms	34,449	1.48771	1.06401	0	10
	Nbathrooms	Number of bathrooms	34,449	1.36505	0.73537	0	10
	Airconditioning	Air conditioning (yes = 1)	34,449	0.32924	0.46994	0	1
	Pool	Pool (yes = 1)	34,449	0.04563	0.20869	0	1
Crowdsourcing	Reviews	Average number of reviews per month	34,449	1.8325	1.76143	0.01	14.11
	Star-rating	Average overall star rating	34,449	93.67514	6.7817	20	100
	Sentiment	Sentiment score (1—low; 2—medium; 3—high)	27,262	2.00631	0.81593	1	3

presents detailed descriptions of the variables used in our study.

4. Empirical Results

4.1. Results

In the first stage of our analysis, we estimated an Ordinary Least Squares (OLS) regression, following the approach adopted in prior studies (Chica-Olmo et al., 2020). Furthermore, we conducted three diagnostic tests to investigate the presence of spatial correlations: the Lagrange Multiplier (LM) test for autoregressive conditional heteroskedasticity (ARCH), the Breusch–Godfrey LM test for autocorrelation, and Moran’s I test for spatial dependence. The results of these tests consistently rejected the null hypothesis of no spatial dependence, providing strong evidence of spatial correlations in the dataset. These findings are summarized in

Table 3. OLS regression diagnostic tests for spatial dependence.

Test	Chi2
LM (ARCH)	20.251
Breusch–Godfrey LM	45.350
Moran	17.21

All tests have Prob > Chi2 = 0.000.

.

Subsequently, a spatial autoregressive regression (SAR) was estimated in STATA using the generalized spatial two-stage least squares estimator (gs2sls), as this method provides consistent estimators in the presence of spatial dependence (Stata, 2022). Our research examines the effects of a broad set of explanatory variables, with particular emphasis on the roles played by variables related to the Host, Booking, Location, and Listing characteristics.

To investigate the influence of online review-related variables on listing prices, we estimated two separate spatial regression models, the results of which are presented in

Table 4. Hedonic regression results.

Variable	Model 1	Model 2
w_ price	0.01062 **	0.01239 **
	(0.00492)	(0.00514)
lang	0.01856 ***	0.02007 ***
	(0.00285)	(0.003)
gender	0.00588 **	0.00728 ***
	(0.0025)	(0.0026)
host_years	−0.00711 ***	0.00682 ***
	(0.00065)	(0.00067)
resp_time	−0.00301	−0.01167 **
	(0.00408)	(0.00454)
resp_rate	−0.05211 ***	−0.06562 ***
	(0.01188)	(0.01458)
acceptance_rate	−0.03533 ***	−0.04832 ***
	(0.01071)	(0.01669)
sh	0.00589 **	0.02578 ***
	(0.00291)	(0.00279)
list_count	0.00061 ***	0.00033 **
	(0.00014)	(0.00014)
cancelation	0.01914 ***	0.01572 ***
	(0.00264)	(0.00274)
min_nights	0.00081 **	−0.00352 ***
	(0.00032)	(0.00054)
is_location_exact	0.0302 ***	0.02766 ***
	(0.00296)	(0.0031)
security_deposit	0.00006 ***	0.00007 ***
	(0.00001)	(0.00001)
cleaning_fee	0.00012	0.0003 ***
	(0.00008)	(0.00009)
guests_included	0.00761 ***	0.01004 ***
	(0.00109)	(0.00114)
extra_people	−0.00014	−0.00052 ***
	(0.00014)	(0.00014)
instant_bookable	−0.00179	−0.00074
	(0.00296)	(0.00308)
quarter	0.00743 ***	0.00723 ***
	(0.00112)	(0.00117)
porto	0.04531 ***	0.05057 ***
	(0.00556)	(0.00624)
neighbor	0.00511	0.02015 ***
	(0.00546)	(0.00618)
dist_station	−0.01358 ***	−0.01853 ***
	(0.00305)	(0.00325)
dist_poi	−0.06122 ***	−0.0641 ***
	(0.00331)	(0.00343)
dist_coastline	−0.06751 ***	−0.06805 ***
	(0.00336)	(0.00367)
accommodates	0.02204 ***	0.01885 ***
	(0.0011)	(0.00117)
room_type	0.19572 ***	0.20479 ***
	(0.00382)	(0.00405)
nbedrooms	0.03751 ***	0.03969 ***
	(0.00223)	(0.00235)
nbathrooms	0.03577 ***	0.04361 ***
	(0.00243)	(0.00258)
airconditioning	0.0621 ***	0.05985 ***
	(0.00286)	(0.00294)
pool	0.18547 ***	0.17014 ***
	(0.0065)	(0.00717)
reviews	−0.02349 ***	−0.02124 ***
	(0.00079)	(0.00082)
ovrating	0.0035 ***
	(0.00024)
sentiment		0.00884 ***
		(0.00168)
_cons	1.49964 ***	1.84138 ***
	(0.03093)	(0.02687)
Pseudo R2	0.50536	0.52573

Legend: ** and *** means p < 0.01, and p < 0.001.

. The first model (Model 1) includes only the overall quantitative ratings assigned by customers, whereas the second model (Model 2) incorporates the sentiment scores extracted from textual reviews.

Multicollinearity is often a concern in hedonic pricing models due to the potential correlation among explanatory variables. To assess the severity of multicollinearity, we computed the Variance Inflation Factor (VIF) for each independent variable. Following the guidelines established by Kennedy (1985), a VIF value exceeding 10 is indicative of problematic multicollinearity. In our analysis, the highest VIF value observed among the independent variables was 4.2, suggesting that multicollinearity does not pose a significant issue in our study.

The results presented in Table 4 report the signs and statistical significance of the estimated coefficients. Host-related attributes play a significant role in determining listing prices. Specifically, the host response rate and acceptance rate both exert statistically significant negative effects on listing prices, though the magnitude of their impact varies depending on the model employed. Conversely, composing the listing description in English is associated with higher prices, suggesting that hosts targeting international clientele tend to charge premium rates. Additionally, being a female host is associated with higher prices.

Interestingly, in line with D. Wang and Nicolau (2017), who argue that prices increase with the number of listings managed by a host, we observe a positive relationship between the number of listings and price, a finding particularly pronounced in Model 1, which includes only the overall quantitative rating. However, contrary to the findings of Y. Chen and Xie (2017), our results indicate that shorter response times negatively impact listing prices in Model 2, while higher acceptance and response rates are generally linked to lower prices. Our study also confirms that Superhost status is strongly associated with favorable reviews and higher prices (Airbnb, 2022).

Regarding attributes related to booking procedures, providing an exact geolocation for the listing (i.e., enabling guests to navigate efficiently and safely from transportation hubs to the property) and offering a flexible cancellation policy exert the most substantial positive impacts across the SAR models. It is noteworthy that the minimum number of nights required has a negative effect on price in Model 2, despite its positive effect in Model 1. Seasonal factors and the number of guests included in the base price both contribute positively to price levels across models. Unsurprisingly, higher security deposits and cleaning fees are associated with increased listing prices. It is also noteworthy that allowing instant booking does not exhibit a statistically significant effect on pricing.

With respect to location attributes, we find that properties situated within the municipality of Porto command higher prices across all models, while listings in neighboring municipalities also show price premiums in Model 2. Consistent with Gyódi and Nawaro (2021), our findings indicate that Airbnb prices decline as the distance from metro stations and key points of interest increases. Moreover, in line with Perez-Sanchez et al. (2018), we observe that accommodation prices decrease as the distance from the coastline increases. Among locational variables, the distance to the metro station exhibits the smallest impact relative to distance from the coastline and other points of interest.

As expected, accommodation characteristics exert significant positive effects on listing prices. The most substantial impact relates to the rental policy, specifically whether the property is offered as an entire place versus a single room, highlighting the premium that guests place on privacy. The number of bedrooms, bathrooms, and maximum guest capacity each contribute positively to pricing. Furthermore, amenities, particularly the presence of a swimming pool, significantly increase prices. Consistent with Guttentag et al. (2018) and Guttentag (2016), Superhost status remains a significant positive determinant of price.

Lastly, in examining crowdsourcing-related attributes, our results align with those of Gibbs et al. (2018), indicating that a greater number of reviews is associated with lower prices. While the overall quantitative rating positively influences listing prices in Model 1, we find that the sentiment score extracted from textual reviews exerts a stronger positive effect in Model 2, contributing to a higher pseudo R-squared value. A closer examination of the quality-signaling variables reveals that their indirect effects exhibit similar patterns across models, underscoring the consistency of their influence on pricing dynamics.

Table 5. Decomposition of the direct and indirect effects of quality variables.

SAR Model	Quality Variable	Direct	Indirect	Total
1	Overall Rating	0.0035023	0.0000010	0.0035033
2	Sentiment Score	0.0086416	0.0000029	0.0086445

shows that direct effects predominate for both the overall rating and the sentiment score, suggesting that listing prices are mostly determined “locally”—changes in a unit’s predictors primarily affect that unit’s outcome, without meaningfully influencing neighboring units. In the context of Airbnb pricing, the small indirect effects imply that amenities or features of one listing, as might be expected to some extent, do not substantially affect the prices of nearby listings.

4.2. Discussion of the Results

Our results are consistent with the broader literature on the determinants of Airbnb listing prices. Specifically, we find that hosts managing multiple listings tend to set higher prices, in line with Abrate and Viglia (2022), Cai et al. (2019), and Gibbs et al. (2018), likely reflecting economies of scale and operational efficiencies achieved through experience and portfolio management. Female hosts are more successful in obtaining higher prices, as found out by Marchenko (2019), which suggests that women convey a sense of trust, interpersonal skills, and safety. In line with Y. Chen and Xie (2017), the response time and acceptance rate exert a negative influence on prices, while the impact of the Superhost status operates in the opposite direction, in line with the conclusions of Airbnb (2022), Guttentag et al. (2018), and Guttentag (2016). Hosts offering lower-priced listings often rely on high occupancy to maximize revenues and to cover operational costs. To attain this objective, hosts may respond very quickly to inquiries in order to secure bookings before the inquirer and potential guests choose another low-cost competitors.

Moreover, we observe that a higher number of reviews is associated with lower prices, in line with Gibbs et al. (2018), Gyódi and Nawaro (2021), and Lin and Yang (2023), reflecting the tendency for relatively inexpensive properties to attract a larger volume of bookings and, consequently, more reviews, albeit at lower price points. The impact of property characteristics is in line with other findings reported in the literature.

Consistent with Zervas et al. (2021) and Östh et al. (2025, p. 7), our study also finds that booking restrictions, such as minimum stay requirements, are associated with more affordable prices, suggesting that listings imposing stricter conditions may need to compensate by offering lower rates to remain competitive. The negative correlation between pricing and the number of guest reviews further underscores the importance of customer feedback in shaping listing desirability and hosts’ pricing strategies.

As anticipated, and in alignment with Lin and Yang (2023), D. Wang and Nicolau (2017), and Jiao and Bai (2020), our findings indicate a negative relationship between listing prices and the distance to key geographic features, including metro stations, cultural points of interest, and the coastline (Lin & Yang, 2023; Zhang et al., 2017; Shokoohyar et al., 2020). Proximity to the city center—proxied by the variable indicating the municipality of Porto—emerges as a significant determinant, positively correlated with listing prices, highlighting the critical role of central locations near major attractions in influencing pricing strategies. In line with Gunter and Önder (2017), this study found that a greater distance from the city center and other areas of touristic interest is a significant negative driver of Airbnb demand. As in Vienna, listings located farther from the city and/or access to public transportations (proxied by the variable dist-station) showed significantly lower demand and, by implication, tend to have lower prices.

This study advances our understanding of the relative importance of various attributes affecting Airbnb listings in the Porto area, including locational factors. Our empirical findings support initial hypotheses regarding the influence of accommodation characteristics, proximity to key geographic features, and neighborhood attributes, particularly regarding proximity to the central city of Porto. In particular, this study demonstrates the impact of various location-related factors, such as proximity to the city center, access to public transportation, surrounding area characteristics, and proximity to the coastline. The impact of geography extends beyond the city center. Properties located farther from the center but close to the metro or the coastline can, to some extent, offset their less central location.

Despite their inherently subjective nature, online reviews remain an indispensable source of information and cannot be overlooked. They serve as a critical determinant of prospective customers’ decisions, as they encapsulate opinions and affective responses shaped by direct, real, intrinsically personal, and in situ experiences. Such narratives provide valuable insights into the anticipated user experience. However, online reviews are intrinsically shaped by the reviewers’ personal histories, expectations, and prevailing emotional states at the precise moment they dedicate time. Sentiments, experiences, and emotions are strong predictors of price and should be carefully managed by Airbnb proprietors and managers, as found out by Lawani et al. (2019), Martinez et al. (2017), and Östh et al. (2025).

While our dataset extends only through 2020, the relatively limited recency of the data is not a major concern, given that investments in the Airbnb sector often involve long-term property enhancements and renovations. Insights derived from 2020 data remain highly relevant, particularly for properties currently undergoing renovation or strategic upgrades. In such contexts, decisions regarding features like the addition of a swimming pool or other premium amenities continue to be informed by the market dynamics and pricing determinants highlighted in this study.

5. Conclusions, Limitations, and Future Work

The regression analysis conducted in this study reveals the multifaceted nature of the determinants of Airbnb listing prices, highlighting the significant influence of host and listing characteristics, as well as the critical impact of the affective component of online reviews along with geographical factors. These findings offer valuable insights for hosts, guests, and the broader hospitality industry. This study contributes to the analysis of price determinants by examining the impact of sentiment scores derived from guest reviews as a proxy for the “quality of the experience.” It contributes meaningfully to the ongoing debate about the attributes influencing Airbnb listing prices, particularly emphasizing the importance of delivering memorable experiences that generate positive online reviews.

It is widely acknowledged that online reviews decisively influence consumers’ purchase decisions. However, quantitative ratings alone may yield biased estimates of experience quality, as they fail to reflect the nuanced sentiments and detailed opinions expressed by guests. Thus, this article contributes to the literature exploring the impact of memorable experiences on positive online reviews and their effect on pricing in the tourism industry, by analyzing reviews through sentiment scores to assess their influence on price. To the authors’ knowledge, this research perspective has received relatively limited coverage. Our findings demonstrate that sentiments extracted from reviews serve as a better proxy for perceived quality than aggregate numerical ratings.

In this study, we confirm that host-related attributes and listing characteristics are crucial determinants of listing prices. Functionality-related factors, such as property size, number of bedrooms and bathrooms, and the availability of amenities, dominate consumers’ evaluations of Airbnb accommodations. Our study also provides deeper insights into the role of location in influencing prices. Properties situated in central urban areas command higher prices than those in surrounding municipalities. Additionally, geographically related amenities in the vicinity, such as proximity to metro stations, cultural points of interest, and coastal areas, significantly influence price levels. This study highlights the influence of a wide range of factors—some previously examined and others analyzed specifically in this research—on the determination of Airbnb prices. Property owners can derive valuable insights, particularly regarding the impact of various amenities on pricing, the advantages of a central location or proximity to the metro (which can reduce the need for car rental services), and the extent to which investing in enhancing one’s hosting skills may be worthwhile. The hypotheses advanced in this study were therefore supported and corroborated. Higher levels of positive sentiment, as reflected in encouraging online reviews, are associated with higher prices, thereby signaling to property owners the need to invest in delivering experiences aligned with customer expectations. Geographical factors were also found to influence prices, consistent with previous research. Furthermore, a greater level of amenities increases customers’ willingness to pay, while highly experienced and professional hosts can expect to command price premiums.

Our research yields two important managerial implications for Airbnb hosts and investors. First, it underscores the importance of analyzing customer reviews beyond mere numerical ratings as a more reliable measure of perceived quality. Second, from an investment perspective, it highlights the strategic significance of the location and surrounding amenities in achieving higher listing prices. Most amenities are positively evaluated by guests, suggesting that targeted investments in property features and neighborhood enhancements can enhance both guest satisfaction and revenue potential.

Some additional managerial implications arise from this study. Properties located in the city center and surrounding areas, as well as in neighboring municipalities, command a price premium. Likewise, properties close to metro stations, coastal areas, and tourist attractions also command higher prices. These results are consistent with the main conclusions of earlier research but offer a degree of novelty for understanding the complex factors driving Airbnb pricing across the Porto district (Lin & Yang, 2023), particularly the multitude and/or combination of geographical factors that enable the charging of higher prices. Given that Airbnb’s economic appeal continues to rely on low prices, and price is recognized as the core competitive advantage of peer-to-peer (P2P) offerings (Tussyadiah & Pesonen, 2016; Gibbs et al., 2018), being frequently cited as the primary reason for choosing Airbnb over traditional hotels (J. Kim et al., 2025; Tussyadiah & Pesonen, 2016; Tussyadiah & Pesonen, 2016), based on this study, proprietors have information to understand which investments should be prioritized to achieve an optimal balance between tourists’ expectations regarding price and amenities and hosts’ expectations for profitability, which is, in turn, dependent on the cost of the amenities and services offered.

Our study is not without limitations. The analysis is confined to the district of Porto, whereas extending the research to additional cities or regions would improve the generalizability of the findings. Furthermore, future research could benefit from incorporating socio-demographic information about reviewers, including nationality, gender, age, and education, to better understand the heterogeneous nature of guest preferences. Expanding the analysis to other geographic contexts and employing time-series data would enable researchers to assess the impact of external shocks, such as the COVID-19 pandemic, on the resilience and evolution of the Airbnb market (Milone et al., 2023). Machine learning models, such as LASSO, could also be explored to identify the most influential variables. Additional predictors, such as the number of listings in the area, could be incorporated into the model. The chosen model (e.g., SAR) assumes a specific functional form and spatial weight structure; as the spatial weight matrix (W) may not perfectly capture real-world neighbor interactions, alternative specifications might yield different results. Furthermore, some potentially relevant factors—such as the number of local events, crime rates, travel time to the airport and to the city of Porto, or dynamic demand patterns—were not included due to data unavailability.

This paper highlights—and corroborates previous literature on—the instrumental role of sentiment in operationalizing online reviews as a management tool. Rather than relying solely on ratings, both customers and managers are increasingly influenced by the emotions and feelings associated with in situ experiences, critical incidents, and both major and minor events. The growing availability of sophisticated yet accessible analytical techniques provides managers with essential tools to assess and evaluate subsets of opinions and comments expressed online. In the post-COVID era, safety concerns and the desire for privacy and social distancing have become increasingly significant in shaping travel behavior. The COVID-19 pandemic underscored the importance of rapid adaptability in a volatile environment, highlighting the value placed on attributes such as property size, number of bedrooms, and the availability of private spaces (Más-Ferrando et al., 2024). Notably, the importance of factors like size and number of bedrooms was already evident in the pre-COVID context, suggesting continuity in consumer preferences across time periods.

This paper highlights and corroborates previous literature, particularly the instrumental role of sentiment in operationalizing online reviews as a management tool. Rather than relying solely on quantitative ratings, both customers and managers are increasingly influenced by the emotions and feelings associated with in situ experiences, critical incidents, and both major and minor events—an important theoretical conclusion to be drawn from this study. The growing availability of sophisticated analytical techniques provides managers with essential tools to assess and evaluate, in terms of emotions and sentiments, and subsets of opinions and comments expressed online, with respect to a specific period of interest. While the market in the Porto region remains open to new business ventures, the increasingly negative perception of the sector among residents should prompt managers to implement measures aimed at mitigating the adverse effects of the large influx of temporary residents, while fostering the social and economic involvement of the local community to demonstrate the benefits of tourism. The sector’s success has resulted in hundreds, if not thousands, of houses and flats being removed from the rental market for local residents. In certain urban areas, according to media reports, only a minority of locally born inhabitants remain, as they are unable to afford the inflated housing prices. Conversely, the Airbnb sector often serves as a last-resort source of income for many middle-aged ‘entrepreneurs’ who face limited employment opportunities elsewhere. Implementing additional taxation could help maintain—rather than expand—the current scale of operations, while financing a dedicated fund to support the development of state-subsidized or cost-controlled housing aimed at enabling low- and middle-income families to move up the housing ladder.

In conclusion, this study enriches the understanding of the complex factors influencing Airbnb pricing, revealing how both tangible attributes and subjective guest sentiments shape market dynamics. As the short-term rental sector continues to evolve in a competitive and digitally mediated environment, incorporating sentiment analysis alongside traditional hedonic modeling offers a powerful tool for hosts, investors, and policymakers to navigate pricing strategies, enhance customer experiences, and ensure sustainable growth in the sharing economy.