Zooming into Airbnb listings of European cities: Further investigation of the sector’s competitiveness

Abstract

Airbnb has a major role to play in the competitiveness of the overall accommodation sector of individual destinations and it is rather unlikely that this role will diminish in the post-COVID-19 recovery of the tourism industry. Therefore, the present study motivates the Airbnb sector to look back at its past performance for insights that can be used in setting post-pandemic targets. In particular, this research assesses competitiveness of the Airbnb listings of 28 European cities by including hotel-related data as uncontrollable input variables within interactive data envelopment analysis modeling. The contribution lies in joining Airbnb listings and hotels into the benchmarking discussion and efficiency analysis, along with looking beyond the cumulative number of listings by dissecting the overall sector into commercial and private listings—something that has not been attempted as of yet, in spite of the ever-growing body of literature on the sharing economy.

Keywords

Airbnb competitiveness data envelopment analysis efficiency sector benchmarking uncontrollable variables

Introduction

In 2020, the COVID-19 pandemic has brought the entire tourism and hospitality industry to its knees, resulting in a decrease in international tourist arrivals by 74% compared to 2019 (UNWTO, 2021). In spite of a cautious outlook recognizing many negative impacts of the pandemic, an eventual industry recovery is unquestionable due to its economic significance and in the light of the vaccine rollout that is under way (UNWTO, 2021). Accordingly, it is appropriate for various tourism sectors to consider their figures from pre-COVID-19 times as a basis for setting targets for the post-COVID-19 recovery. Such an approach is also already practiced at the destination level (see e.g., ECM, 2020).

The sharing economy, especially Airbnb, has had a major impact on the accommodation sector over the last few years. More often than not, negative impacts have been highlighted. Dolnicar and Zare (2020) describe Airbnb as a “disruptor” which is now being “disrupted” by the pandemic. Different scenarios have been hypothesized regarding the impacts of COVID-19 on the future of Airbnb (e.g., drop in investor-hosted Airbnb listings or recovery but not to pre-pandemic levels as per Dolnicar and Zare, 2020; short-term shift toward the regular rental market as per Kadi et al., 2020, etc.).

Still, Airbnb has a major role to play in the competitiveness of the overall accommodation sector of individual destinations and it is rather unlikely that this role will diminish in the post-COVID-19 recovery of the tourism industry. Therefore, inspection of efficiency of the Airbnb listings remains a topic of interest, as it can be presupposed that the ultimate goal of Airbnb hosts—but also of regulatory bodies—is to make the Airbnb sector more efficient, i.e., competitive. The ongoing pandemic provides the opportunity to look back at the past performance of this sector to gain insights that can be used in setting post-pandemic targets.

With the above in mind, the present study builds upon the research by Zekan et al. (2019) by assessing the competitiveness of the Airbnb listings of 28 European cities. In more detail, this study aims at including uncontrollable input variables relating to traditional forms of accommodation establishments, specifically hotels, in modeling the sector’s competitiveness. This has not been attempted in the Airbnb competitiveness studies conducted to date and is done in order to further account for heterogeneity. Data envelopment analysis (DEA), a method which is often praised for its superior benchmarking abilities and allows modeling of uncontrollable variables, has been selected for calculation of efficiency cores. Besides investigating the efficiency of an average listing per city, an additional layer of DEA runs is added by separating private and commercial listings. Thus, this study not only inspects the “big” picture (cumulative Airbnb listings), but also seeks to reveal any differences in efficiency between the two clusters. Such investigation constitutes another contribution of the present research.

In brief, the findings present an overview of efficient and inefficient listings of European cities across various models, which demonstrate the impact of different variables on the efficiency scores. The potential for improvement is identified for all inefficient Airbnb listings. Another merit of this benchmarking endeavor is the proposal of individual best practices. Moreover, this is the first study that joins Airbnb listings and hotels into the benchmarking discussion and efficiency analysis and as such is of interest to both tourism researchers and tourism practitioners.

The remainder of this article is organized as follows. The second section reviews the related literature on Airbnb, while the third section provides insights into the economic theory in line with DEA, model, and data. Empirical findings and discussion follow in the fourth section. The fifth and concluding section highlights the most interesting takeaways, along with limitations and suggestions for future research.

Related literature

A quick Google Scholar search performed on 10 March 2021, yielded approximately 3,670,000 results for the search term “sharing economy,” underling the continuing academic interest in this research topic. Belk (2007) defines sharing as follows:

“Sharing is an alternative to the private ownership that is emphasized in both marketplace exchange and gift giving. In sharing, two or more people may enjoy the benefits (or costs) that flow from possessing a thing. Rather than distinguishing what is mine and yours, sharing defines something as ours.” (Belk, 2007: 127)

While sharing as such is not a novel phenomenon between relatives, friends, or neighbors, Web 2.0 technologies have enabled sharing between complete strangers using peer-to-peer online platforms as confidence builders (Grassmuck, 2012; Green, 2012; Hawlitschek et al., 2016). Wirtz et al. (2019) suggest using the term sharing economy as an umbrella term for oftentimes synonymously employed terms such as collaborative consumption, gig economy, or peer-to-peer economy. Said peer-to-peer online platforms can also be interpreted as two-sided marketplaces characterized by same-sided and cross-sided positive network externalities. The platforms thus operate as intermediaries which reduce search and transaction costs (see, e.g., Parker and Van Alstyne, 2005, or Rochet and Tirole, 2006, for further details). As the platforms providing the highest network externalities are typically preferred by market participants, a tendency toward market concentration with only few intermediaries (or even only one intermediary characterized by a high degree of monopoly power) is not uncommon.

As the sharing economy also includes commercial peer-to-peer platforms (Belk, 2014), Oskam and Boswijk (2016) and later Oskam (2019) argue that the commercial aspect of a non-negligible number of these platforms makes the term “sharing” somewhat misleading. Moreover, the relationship between the “peers” of such platforms may not always be symmetric as not everybody has the financial means to purchase items, real estate, etc., before sharing it with others (Gunter, 2018). Ranchordás (2015) underlines that the lack of adequate regulation for sharing economy providers, customers, and employees alike (Quattrone et al., 2016; Schor, 2016) may result in unfair competition between market participants.

The sharing economy has entered and transformed numerous industries and enabled the development of still more. These range from crowdfunding (e.g., Belleflamme et al., 2014; Mollick, 2014), crowdsourcing (e.g., Difallah et al., 2015; Thebault-Spieker et al., 2015), decision support systems (e.g., Bolton et al., 2013; Vavilis et al., 2014), and redistribution markets (e.g., Cabral and Li, 2015; Haucap and Heimeshoff, 2014), to enabling ride and vehicle sharing (e.g., Parkes et al., 2013; Santi et al., 2014). Not too surprisingly, the sharing economy has also successfully entered the accommodation industry. The commercial platform Airbnb (www.airbnb.com) has become somewhat of an eponym for the sharing economy in the accommodation industry, even though other peer-to-peer online platforms such as Couchsurfing (www.couchsurfing.com) exist in the accommodation industry, albeit with a somewhat different customer base and business model than Airbnb (Molz, 2013). Airbnb has come a long way from its foundation based on the idea of an “airbed and breakfast” in San Francisco in 2007 (Choe, 2007) to the dominant market position it holds today.

Reportedly, the COVID-19 pandemic and its associated travel and accommodation restrictions have affected Airbnb substantially: as of November 2020, its revenue had declined by 72% and 1,800 employees had been laid off since the outbreak of the pandemic (Abril, 2020). Still, Airbnb counts 5.6 million active listings in 100,000 cities across more than 220 countries and regions as of 30 September 2020 (Airbnb, 2021). On 10 December 2020, Airbnb successfully went public on Nasdaq, now being worth more than the three biggest hotel chains combined (Sonnemaker, 2020). As with the rest of the accommodation industry, its business is therefore likely to recover once the pandemic is over. Guttentag (2015) describes Airbnb’s business model as a disruptive innovation, i.e., one that has been able to change the market structure of the accommodation industry, very much in line with Schumpeter’s (1942) earlier notion on creative destruction. The author also notes the downsides of this development, notably the danger of “touristification” of residential areas (i.e., local businesses starting to cater to the needs of tourists rather than residents, rising rent and overall price levels, amplifying housing shortages, etc.) and Airbnb providers not complying with (local) regulations (Guttentag, 2015).

Moreover, not only the fee-commissioning platform itself can be characterized as commercial, but also a non-negligible portion of its providers, i.e., commercial Airbnb providers, who offer more than one listing on the platform (e.g., Messe Berlin, 2016). While Ert et al. (2016) find that Airbnb providers perceived as trustworthy according to their profile pictures are able to charge higher prices and are more likely to receive booking inquiries, non-Caucasian Airbnb providers do not seem to benefit from such assessments of trustworthiness (Edelman and Luca, 2014). Also booking inquiries from non-Caucasian (potential) Airbnb customers seem to be less successful (Edelman et al., 2017). Altogether, Airbnb has drawn so much attention from researchers that multiple literature reviews on Airbnb and the sharing economy in the accommodation industry have been published (in the English language) to date. The main findings of seven recent such meta studies are briefly summarized in the following.

Reviewing 118 published journal articles on Airbnb, Dann et al. (2019) find that Airbnb research can be grouped into seven categories: (1) motivation for its use, (2) trust and reputation, (3) reviews, (4) role of photographs, (5) pricing, (6) impact on the accommodation industry, and (7) regulatory aspects. According to the authors, the reviewed studies mostly use methods suitable for surveys and secondary data. Guttentag (2019) reviews 132 journal articles on the topic and concludes that most of them have a European or North American geographical focus and are published in tourism and hospitality journals. The author groups the reviewed studies into six categories, notably: (1) guests (or customers), (2) hosts (or providers), (3) impacts of Airbnb supply on destinations, (4) regulatory aspects, (5) impacts of Airbnb on the tourism sector, and (6) the Airbnb platform itself. Kuhzady et al. (2020) review 371 published journal articles on the sharing economy in the accommodation industry with a focus on Airbnb concluding that (1) authenticity, (2) environmental issues, and (3) the possibility of socialization between providers and customers have been important topics for researchers.

With a more general sharing economy approach, Belarmino and Koh (2020) explore 107 published articles and consider the topics of (1) affordable housing, (2) conceptualization of the sharing economy, (3) consumer behavior, (4) regulatory aspects, (5) relationship between the sharing economy and hotels, (6) revenue management, as well as (7) trust and mistrust as prevalent in research. Hossain (2020), in turn, analyzes the findings of 219 journal articles, highlighting the sharing economy’s novel business models and revenue streams on the one hand and regulatory and policy deficits on the other. In the same vein, Sainaghi (2020) reviews 189 published journal articles and discovers the subsequent eight research topics: (1) peer-to-peer platforms, (2) demand-side studies, (3) hosts (performance, pricing, and location), (4) guest and host relationships, (5) economic impact, (6) social impact, (7) regulatory aspects, and (8) the sharing economy itself as a phenomenon. Reviewing 305 articles, Hodak and Krajinović (2020) conclude that overtourism has been an under-researched topic in the sharing economy literature to date.

Despite the differences in the classifications of the above systematic literature review studies, one common observation from analyzing the literature on Airbnb is researchers’ interest in the relationship of Airbnb/the sharing economy and the traditional actors of the accommodation industry, notably hotels. For instance, Airbnb can be viewed as a positive innovation in those countries without sufficient hotel room supply (Adamiak, 2018). Gunter and Önder (2018) conclude that Airbnb should be seen as a complement to hotels in Vienna given the lack of sought-after hotel room types comparable to Airbnb listings. Leick et al. (2021) find a positive long-run effect of Airbnb demand on traditional accommodation in secondary Norwegian destinations, thus also supporting the notion of a complementary relationship. Sainaghi and Baggio (2020) cannot confirm a postulated substitutional nature of Airbnb for the Italian city of Milan because of opposing seasonal patterns. In a different study, the same authors conclude for the cities of Milan and Rome that also the demand for commercial Airbnb listings features different seasonal patterns than that for hotels in both cities (Sainaghi and Baggio, 2021).

For New York City, however, Gunter et al. (2020) find that hotels and Airbnb accommodation should be treated as substitutes. Farronato and Fradkin (2018) argue that for New York City and similar destinations, one reason for this substitutional nature is capacity constraints of hotels, particularly during peak seasons. Evaluating an online survey of more than 800 Airbnb customers, Guttentag and Smith (2017) conclude that more than two thirds of the survey respondents see Airbnb as a substitute to hotels. In general, Airbnb can still be considered as the less expensive accommodation option, while often being located close to major points of interest, notably in city destinations (e.g., Adamiak et al., 2019; Eugenio-Martin et al., 2019; Gunter, 2018; Önder et al., 2019; Tong and Gunter, 2020). Using a difference-in-differences approach, Falk and Yang (2020) find that stricter regulations of Airbnb listings and their providers benefit hotels in five European cities (Amsterdam, Barcelona, Berlin, London, and Paris). Employing the same approach to New York City and Washington, DC, Yeon et al. (2020) find beneficial effects only for low-cost hotels. For Houston, Texas, only a moderate substitutional relationship between Airbnb and hotels can be confirmed (Yang and Mao, 2020). While also Zervas et al. (2017) see Airbnb mostly as a threat to traditional bed and breakfasts and low-cost hotels, Dogru et al. (2019, 2020) conclude that Airbnb has become a relevant competitor to hotels across hotel classes in terms of typical performance measures for a number of countries.

Nonetheless, the literature on competitiveness of Airbnb (in comparison to hotels) is still scarce. While there are very few publications integrating the Airbnb phenomenon into broader destination competitiveness studies (e.g., Koo et al., 2016; Skalska and Shcherbiak, 2016), to the best of the authors’ knowledge, there is only one study investigating the competitiveness of Airbnb listings as decision making units: Zekan et al. (2019). Still lacking in the existing literature, however, is an investigation of the impact of hotel-specific variables on the competitiveness of Airbnb listings, as well as a separate investigation of commercial and private Airbnb listings, whereby the former may ex-ante be characterized by a more efficient use of their inputs than the latter. This is the gap in the literature the present study tries to fill, which has also only very recently been confirmed as a research gap in a systematic literature review on the competitiveness of the visitor economy by Kim et al. (2021).

Methodology

In neoclassical production theory, firms are assumed to use their scarce inputs efficiently to produce the desired output(s) (e.g., Rasmussen, 2013; Sickles and Zelenyuk, 2019). More traditional approaches in both macro- and microeconomics employ different types of production functions to model production technology, i.e., the output variable as a function of one or more input variables. When efficient production based on all inputs and all outputs of all firms in an industry (or in the economy as a whole) is evaluated jointly, the aggregation of all individual production functions results in a production possibility frontier, at least in theory. Even though some types of production functions, such as the constant elasticity of substitution (CES) production function (Arrow et al., 1961; Solow, 1956), can feature the realistic property of variable returns to scale (i.e., the possibility of possessing constant, increasing, or decreasing returns to scale), researchers have criticized such parametric functional forms for a number of reasons. In aggregate terms, in particular their empirical observability and relevance since stemming from a national accounting identity has been subject to criticism (e.g., Cohen and Harcourt, 2003; Daly, 1997; Shaikh, 1974).

In their pivotal study, Charnes et al. (1978) therefore suggest a new approach to obtain non-parametric production functions and production possibility frontiers (or rather: best practice frontier) that are directly obtained from observed sample data: the DEA approach, which itself is based on the earlier work of Farrell (1957). As DEA efficiency scores are obtained exclusively from observed sample data without the need for knowledge about the asymptotic properties of a theoretical population (e.g., theoretically possible levels of efficiency), only relative efficiency of peer entities can be measured. In DEA, these comparative entities are referred to as decision making units (DMUs) (Cooper et al., 2004).

As their originally suggested DEA model, the Charnes–Cooper–Rhodes (CCR) or constant returns to scale (CRS) model, still suffers from the namesake constant returns to scale property, a later refinement, the Banker–Charnes–Cooper (BCC) or variable returns to scale (VRS) model (Banker et al., 1984), relaxes this assumption and is therefore employed hereinafter (Cooper et al., 2004). Full proportionality between input and output variables is not presupposed (Assaf and Agbola, 2011). Moreover, this particular DEA model is deemed appropriate for situations when not all DMUs are operating at an optimal scale, for instance, due to financial constraints or imperfect competition that are likely to exist in the accommodation industry (Assaf and Agbola, 2011; Wöber, 2002).

Next, the model orientation needs to be considered. The output-oriented model aims at becoming efficient by increasing outputs and holding inputs fixed, unlike the input-oriented model that focuses on input reduction and fixed outputs. There is also a non-oriented or base model with a simultaneous improvement of both inputs and outputs (Donthu et al., 2005; Scheel, 2000). As in the present study, outputs are sought to be maximized based on an efficient use of existing inputs, the output-oriented BCC model becomes relevant. Mathematically, it is a constrained maximization problem and reads as follows (Cooper et al., 2004):

\max φ + ε (\sum_{i = 1}^{m} s_{i}^{-} + \sum_{r = 1}^{s} s_{r}^{+})

subject to:

\sum_{j = 1}^{n} x_{i j} λ_{j} + s_{i}^{-} = x_{i 0}, i = 1,2, \dots, m;

\sum_{j = 1}^{n} y_{r j} λ_{j} - s_{r}^{+} = φ y_{r 0}, r = 1,2, \dots, s;

λ_{j} \geq 0, j = 1,2, \dots, n;

\sum_{j = 1}^{n} λ_{j} = 1

Here, $φ$ denotes the efficiency score, $ε$ a non-Archimedean element that is smaller than any positive real number, $s_{i}^{-}$ and $s_{r}^{+}$ two slack variables, $y_{r 0}$ and $x_{i 0}$ the observed outputs and inputs, and $λ_{j}$ the corresponding weights (see Cooper et al., 2004, for a more detailed explanation).

The standard model formulations oftentimes presuppose “standard” variables, which can be controlled by DMUs. However, there may be other variables that need to be considered: those beyond their control, yet with impact on performances of DMUs. Such variables are described as non-discretionary or uncontrollable and can be modeled in DEA alongside discretionary or controllable variables, as is shown, for instance, in Cooper et al. (2004), Reynolds (2003), Tsai et al. (2011), Wöber (2002), and Wöber and Fesenmaier (2004). In the end, it all comes down to ensuring that all relevant variables are included as, otherwise, the entire investigation will be of little value (Cook et al., 2014). This begs the question regarding the appropriate number of variables in relation to the number of DMUs to preserve the method’s discriminatory power. As a response, Cook et al. (2014) conclude that the suggested “rule of thumb” (number of DMUs = twice/three times inputs + outputs) has no statistical basis and thus advise against a sample size requirement in DEA.

The selection of variables in the present study is grounded in Zekan et al. (2019), who provide an in-depth empirical and theoretical justification for every variable employed in their analyses. In more detail, the empirical justification for the selected controllable inputs is their explanatory power as determinants of Airbnb demand measured, for instance, in terms of occupancy rate from regression analysis (Gunter and Önder, 2018). The underlying theoretical foundation is Airbnb hosts operating in an imperfectly (i.e., monopolistically) competitive industry, where not only price but also (perceived or real) non-price differences of heterogeneous products determine their demand (Chamberlin, 1933; Dixit and Stiglitz, 1977; Edwards, 1933; Robinson, 1933). For the selection of the uncontrollable hotel-related inputs, the competitive relationship between the sharing economy and the traditional actors of the accommodation industry has proven relevant (see the “Related Literature” section). Since the focus lies on investigating competitiveness of the same type of DMU (= average Airbnb listing of a city), the most plausible approach is to model the same types of variables and add uncontrollable inputs into the mix. By doing so, the present research follows the plea for standardization of variable selection by Assaf and Josiassen (2016). Hence, the main DEA model (hereinafter referred to as model 0; Figure 1) consists of six inputs and two outputs. Four controllable inputs are all Airbnb-specific and are as follows: number of listings, maximum number of guests, number of photos, and response rate. Two uncontrollable inputs are hotel-related: ADR and occupancy rate of hotels. Two outputs are ADR and occupancy rate of Airbnb. As is detailed in the “Findings” section, some of the variables of the main model have been replaced with four new variables in the interactive DEA (models 1–4): number of bedrooms, minimum number of nights, census rooms hotels, and number of bookings. Overall, it should be underlined that in every model there are variables that address both Airbnb demand and supply.

Figure 1.

The main data envelopment analysis model (model 0).

The Airbnb-specific data were obtained from AirDNA (https://airdna.co), whereas hotel data were generously provided by STR (https://str.com). This study covers the period between July 2015 and June 2016 and considers only Airbnb listings that were active (i.e., non-zero bookings) during the aforementioned period. With exception of the number of listings, all variables are averages of active listings per city, representing the city’s Airbnb sector; hotel-specific variables are also averages. The same approach is followed in the DEA modeling of the cumulative sample, as well as in separate commercial (i.e., listings by hosts that own two or more properties) and private (i.e., listings by hosts who own a single property) samples of 28 cities. The aforementioned definitions of “commercial” and “private” listings have been adopted from Gunter et al. (2020). The list of cities whose Airbnb sector is evaluated along with their commercial/private ratio can be found in Table 1. Efficiency Measurement System (EMS; http://www.holger-scheel.de/ems/) version 1.3 was used for DEA modeling, whereas IBM SPSS Statistics version 26 was used for a non-parametric Wilcoxon test.

Table 1.

Commercial/private ratio of active Airbnb listings per city.

	Percentage (commercial)	Percentage (private)
Amsterdam	34.12	65.88
Barcelona	66.63	33.37
Berlin	36.33	63.67
Brussels	48.16	51.84
Budapest	61.74	38.26
Cologne	31.80	68.20
Copenhagen	19.92	80.08
Dublin	51.32	48.68
Edinburgh	49.32	50.68
Florence	66.97	33.03
Geneva	38.27	61.73
Genoa	47.60	52.40
Hamburg	29.27	70.73
Helsinki	24.39	75.61
Istanbul	72.33	27.67
Lisbon	70.01	29.99
London	52.93	47.07
Madrid	59.60	40.40
Munich	31.91	68.09
Oslo	28.19	71.81
Paris	29.72	70.28
Prague	68.00	32.00
Rome	69.27	30.73
Salzburg	43.63	56.37
Seville	61.60	38.40
Stockholm	17.79	82.21
Tallinn	51.82	48.18
Vienna	45.21	54.79

Findings and Discussion

Prior to introducing the efficiency scores, a brief observation can be made regarding descriptive statistics of the eight variables of the main model (Table 2). As is expected due to the varying sizes of DMUs, heterogeneity is present. This holds for the cumulative sample, as well as for the separate commercial and private samples. As this study is concerned solely with the sector’s performance (i.e., organization benchmarking), the observed heterogeneity does not pose an issue in the interpretation of efficiency scores. Nonetheless, it is still accounted for in the benchmarking analyses by including uncontrollable inputs into the DEA modeling and by computing a virtual benchmark. Efficiency of all listings is discussed next.

Table 2.

Descriptive statistics—variables of the main DEA model (model 0).

	Maximum	Minimum	Mean	Standard deviation
Inputs
Number of listings
All listings	66184	612	13132.07	15599.77
Commercial	29581	267	6083.50	6996.31
Private	46513	345	7048.57	9777.70
Maximum number of guests
All listings	4.33	2.64	3.28	0.43
Commercial	4.61	2.76	3.50	0.43
Private	3.89	2.49	3.05	0.37
Number of photos
All listings	23.08	10.18	15.13	3.37
Commercial	23.62	11.37	16.35	3.18
Private	21.83	9.63	13.93	3.18
Response rate
All listings	95.35	87.41	91.84	2.52
Commercial	95.86	88.37	92.79	2.19
Private	94.98	86.74	90.89	2.57
ADR hotels	278.51	76.48	133.36	48.40
Occupancy rate hotels	82.70	56.20	73.10	5.79
Outputs
ADR
All listings	159.15	60.53	99.62	26.39
Commercial	164.34	63.75	106.08	27.88
Private	156.47	55.35	92.31	27.06
Occupancy rate
All listings	59.30	30.33	48.77	6.35
Commercial	57.41	30.72	47.57	6.38
Private	60.45	29.29	49.71	6.41

DEA: data envelopment analysis.

Efficiency Insights Part 1: All Listings

The first level of analysis inspects the efficiency of the cumulative sample. DEA efficiency scores for 28 DMUs range from 80.11% (listings of Amsterdam) to 117.82% (listings of Florence). In the interest of simplification, only city names are displayed in Table 3 and they serve as proxies for the cities’ average Airbnb listing. Out of 28 analyzed DMUs, 17 (60.71%) are efficient, whereas 11 (39.29%) are inefficient as per proposed DEA modeling (Figure 1).

Table 3.

DEA results (model 0)—all listings.

	Inefficient DMU	Score	Benchmarking partners and weights
10	Florence	117.82%	1 (0.38) 4 (0.01) 11 (0.22) 14 (0.21) 26 (0.18)
12	Genoa	116.67%	4 (0.08) 14 (0.41) 24 (0.14) 27 (0.38)
23	Rome	109.94%	1 (0.33) 4 (0.48) 11 (0.18)
28	Vienna	109.56%	3 (0.25) 5 (0.01) 14 (0.72) 24 (0.02)
2	Barcelona	109.01%	1 (0.71) 3 (0.04) 14 (0.24)
25	Seville	106.12%	1 (0.11) 5 (0.03) 14 (0.15) 24 (0.08) 27 (0.62)
21	Paris	105.21%	1 (0.03) 7 (0.27) 11 (0.47) 26 (0.24)
20	Oslo	105.13%	1 (0.07) 4 (0.09) 11 (0.08) 14 (0.76)
18	Madrid	104.79%	3 (0.13) 5 (0.11) 14 (0.65) 27 (0.11)
17	London	103.01%	1 (0.71) 11 (0.29)
16	Lisbon	102.15%	3 (0.37) 5 (0.10) 14 (0.53)
	Efficient DMU	Score	Benchmark appearance
7	Copenhagen	99.73%	1
22	Prague	99.59%	0
9	Edinburgh	93.41%	0
8	Dublin	92.96%	0
5	Budapest	90.50%	4
1	Amsterdam	80.11%	7
3	Berlin	big	4
4	Brussels	big	4
6	Cologne	big	0
11	Geneva	big	5
13	Hamburg	big	0
14	Helsinki	big	8
15	Istanbul	big	0
19	Munich	big	0
24	Salzburg	big	3
26	Stockholm	big	2
27	Tallinn	big	3

DEA: data envelopment analysis; DMU: decision making unit.

Looking into the efficient DMUs first (i.e., maximum performance = 100%; also those with “superefficiency” scores < 100% in the output-oriented BCC modeling), the listings of 11 cities are determined to have “big” scores. Such scores are infeasible solutions and denote units with extremely high efficiency (Boljunčić, as cited in Wöber and Fesenmaier, 2004). The same authors mention that any further evaluation of such cases is not possible. Remaining are the listings of six cities with numerical scores: Amsterdam, Budapest, Dublin, Edinburgh, Prague, and Copenhagen. Although all of them are efficient in the utilization of their inputs, Amsterdam outperforms the others; especially Copenhagen by 19.62%. The present study also confirms the findings of Önder et al. (2017) and Zekan et al. (2019) with regard to benchmark appearances of efficient units: being efficient does not equate to being a benchmark (i.e., best practice). There are seven DMUs with zero benchmark appearances. This suggests that they are not considered as benchmarks for any of the inefficient units. The listings of Helsinki and Amsterdam are the most prominent benchmarks, with eight and seven appearances, respectively. Yet, universal best practice for all inefficient units is not apparent in this study either. Given the heterogeneity of the sample, this is also the expected outcome and reflects real-life situations.

Moving on to the inefficient units (i.e., those with scores > 100% in the employed model), it is evident that the level of inefficiency is varied among the listings of eleven cities. The closest to obtaining the maximum efficiency are the listings of Lisbon (2.15%), London (3.01%), and Madrid (4.79%), whereas the furthest from the cut-off point are Rome (9.94%), Genoa (16.67%), and Florence-based Airbnb accommodations (17.82%). Percentages imply the potential that each DMU has regarding improvement of at least one of its modeled outputs (ADR and occupancy rate) with the same level of inputs. These percentages can be found in Table 3 for every inefficient DMU. The score itself is a DMU’s performance estimation. Yet, if performance improvement is desired, then each DMU should look toward its individualized benchmarking partners, which are also computed with DEA and are presented in Table 3. For instance, let us put listings of Florence (the worst performer) in the spotlight. Five benchmarking partners are proposed for this particular DMU: Amsterdam (0.38), Brussels (0.01), Geneva (0.22), Helsinki (0.21), and Stockholm (0.18). The values in the brackets indicate the relative importance (i.e., weight) of each benchmark for Florence and the interpretation of these results is straightforward: the most important benchmark is the average Airbnb listing of Amsterdam (0.38).

With regard to discussion on benchmarking partners for inefficient units, an additional analytical step can be performed in order to overcome the potential problem of incomparability and by doing so, tackling the issue of heterogeneity. A number of DEA studies (e.g., Wöber and Fesenmaier, 2004; Bauernfeind and Mitsche, 2008; Önder et al., 2017; Zekan et al., 2019) advocate calculating a so-called virtual benchmark. Basically, continuing with the previous examination of Florence listings, this can be accomplished by forming a linear combination of the unit’s five proposed benchmarks and their weights in order to show the optimal output targets (Table 4). More specifically, on the input side, three of Florence-based listings’ inputs are higher when compared to the virtual benchmark (maximum number of guests, number of photos, and response rate). Yet the opposite is the case on inspection of both outputs, which explains inefficiency: listings of Florence reach about 85% of its virtual benchmark for ADR and occupancy rate. Thus, a conclusion can be drawn about the DMU’s overall underutilization of inputs and the improvement potential of outputs up to the values of its virtual benchmark (about 18%).

Table 4.

Virtual benchmark for Florence-based listings.

	Florence	Virtual benchmark	Difference
Inputs
Number of listings	9192	9192	0
Maximum number of guests	3.66	2.95	0.71
Number of photos	19.67	13.09	6.58
Response rate	95.33	90.29	5.04
ADR hotels	173.52	173.52	0
Occupancy rate hotels	73.60	73.60	0
Outputs
ADR	114.57	134.99	−20.42
Occupancy rate	47.42	55.87	−8.45

As DEA measures relative efficiency, Zekan et al. (2019) employed a comprehensive interactive modeling with the purpose of inspecting the robustness of their findings. Önder et al. (2017) and Wöber (2002) advocate in favor of such a simulation approach as it provides a better understanding of benchmarking results. The present study follows a similar approach by comparing the efficiency scores of four additional models: specifically, by altering the main model (model 0) on both the input and output sides. With regard to inputs, number of bedrooms (an alternative listing size proxy), minimum number of nights (measure of the minimum length of stay), and census rooms hotels (alternative uncontrollable variable) are introduced into modeling. On the output side, occupancy rate is substituted with number of bookings as another measure of Airbnb demand. The number of variables remained eight (six inputs and two outputs) at times; the sample size of 28 DMUs also remained unchanged in each alteration, thus satisfying the arguable “rule of thumb.” Table 5 gives an overview of the efficiency scores of each model (i.e., of model 0 and four alterations) for each DMU (in alphabetical order).

Table 5.

Interactive DEA results—all listings.

	DMU	DEA score model 0	DEA score model 1	DEA score model 2	DEA score model 3	DEA score model 4
1	Amsterdam	80.11%	80.11%	80.11%	83.80%	80.42%
2	Barcelona	109.01%	109.01%	109.01%	109.01%	116.43%
3	Berlin	big	big	big	big	big
4	Brussels	big	big	big	big	105.32%
5	Budapest	90.50%	90.50%	90.50%	90.50%	85.54%
6	Cologne	big	big	big	big	big
7	Copenhagen	99.73%	99.18%	95.71%	99.73%	97.87%
8	Dublin	92.96%	88.60%	88.60%	95.70%	85.78%
9	Edinburgh	93.41%	84.17%	84.78%	93.41%	88.77%
10	Florence	117.82%	108.33%	108.33%	117.82%	90.27%
11	Geneva	big	big	big	big	big
12	Genoa	116.67%	big	big	116.67%	big
13	Hamburg	big	big	big	big	big
14	Helsinki	big	big	big	big	big
15	Istanbul	big	big	big	big	176.72%
16	Lisbon	102.15%	101.49%	101.49%	102.15%	101.28%
17	London	103.01%	100.05%	100.89%	103.01%	103.01%
18	Madrid	104.79%	104.79%	104.79%	105.06%	115.12%
19	Munich	big	big	81.96%	big	big
20	Oslo	105.13%	105.13%	105.13%	105.13%	125.73%
21	Paris	105.21%	107.46%	105.21%	big	107.38%
22	Prague	99.59%	98.83%	98.83%	99.59%	94.77%
23	Rome	109.94%	96.21%	96.21%	109.94%	109.25%
24	Salzburg	big	big	big	big	big
25	Seville	106.12%	106.12%	106.12%	106.12%	90.99%
26	Stockholm	big	88.62%	big	big	big
27	Tallinn	big	big	big	big	big
28	Vienna	109.56%	109.56%	109.56%	109.56%	134.38%

DEA: data envelopment analysis; DMU: decision making unit.

Model 0 = Inputs: number of listings, maximum number of guests, number of photos, response rate, ADR hotels, occupancy rate hotels. Outputs: ADR, occupancy rate.

Model 1 = Inputs: number of listings, maximum number of guests, minimum number of nights, number of photos, ADR hotels, occupancy rate hotels. Outputs: ADR, occupancy rate.

Model 2 = Inputs: number of listings, maximum number of guests, minimum number of nights, response rate, ADR hotels, occupancy rate hotels. Outputs: ADR, occupancy rate.

Model 3 = Inputs: number of listings, number of bedrooms, number of photos, response rate, ADR hotels, occupancy rate hotels. Outputs: ADR, occupancy rate.

Model 4 = Inputs: number of listings, maximum number of guests, number of photos, response rate, ADR hotels, census rooms hotels. Outputs: ADR, number of bookings.

To sum up the scores of models 1–4, up to ten DMUs are inefficient, with scores ranging from 100.05% (London-based listings) to 176.72% (Istanbul-based listings). The title of worst performer is shared between Florence, Vienna, and Istanbul, depending on the modeling employed, which also demonstrates the importance of interactive over static DEA runs regarding robustness. In line with the main model, the Airbnb listings of Barcelona, Lisbon, London, Madrid, Oslo, and Vienna continue to be inefficient, which sends a very clear message about the as yet unused potential for improvement. Out of 28 DMUs, seven are particularly interesting and are highlighted in Table 5: the listings of Brussels, Florence, Genoa, Istanbul, Paris, Rome, and Seville. These are the DMUs where even a minor change in variables meant a shift from efficient to inefficient classification (or vice versa) in at least one of the models. Based on the modeling employed, each DMU may gain insight into the variable(s) that led to classification as inefficient and, thus, identify the problem area(s) and look toward the practices of their benchmarks for guidance, which is a rather important practical implication. Benchmarking partners for each inefficient unit in the interactive DEA modeling have been identified as well, yet they are not provided due to space constraints (available from the authors upon request).

On the efficient side, there are up to 12 DMUs with extremely high efficiency (i.e., with “big” scores), whereas the numerical scores are in range from 80.11% (Amsterdam-based listings) to 99.73% (Copenhagen-based listings), in line with the main model. The most prominent benchmarks remain the listings of Helsinki (up to eight appearances) and Amsterdam (up to seven appearances), now joined also by Salzburg (eight appearances in model 4). There are up to nine DMUs that have zero benchmark appearances, corroborating the previous conclusions that “efficient” does not presuppose “best practice.”

Efficiency Insights Part 2: Commercial versus Private Listings

The second level of analysis departs from the cumulative sample and compares the efficiency of commercial and private samples. More specifically, the cumulative sample of Airbnb listings of each of 28 cities has now been separated into two clusters: commercial (i.e., listings by hosts that own two or more properties) and private (i.e., listings by hosts who own a single property). For simplicity reasons, they are referred to as commercial and private listings, with city names being used as proxies on this occasion again (Tables 6–8). Thus, the purpose of this additional layer of DEA runs is to gain insights into their efficiency and to identify differences between the two clusters in order to answer questions such as: Is the average commercial listing of Vienna more efficient than the average private listing of Vienna? Are there any differences in allocation of benchmarking partners between the inefficient commercial and private listings? The same model (model 0) is employed as in the case of the cumulative sample, yet with a separate DEA run on each of the two clusters. Efficiency scores, benchmarking partners, and benchmark appearances are presented separately for commercial (Table 6) and private listings (Table 7), whereas Table 8 summarizes the aforementioned efficiency scores for each DMU (in alphabetical order) for ease of comparison.

Table 6.

DEA results (model 0)—commercial listings.

	Inefficient DMU	Score	Benchmarking partners and weights
10	Florence	116.18%	1 (0.53) 11 (0.25) 14 (0.22)
23	Rome	113.47%	1 (0.33) 4 (0.48) 11 (0.18)
2	Barcelona	106.04%	1 (0.71) 3 (0.04) 14 (0.24)
28	Vienna	105.54%	1 (0.08) 3 (0.23) 5 (0.10) 14 (0.45) 24 (0.13)
17	London	100.68%	1 (0.70) 11 (0.18) 21 (0.12)
16	Lisbon	100.28%	3 (0.50) 5 (0.06) 14 (0.43)
	Efficient DMU	Score	Benchmark appearance
18	Madrid	99.11%	0
21	Paris	98.34%	1
25	Seville	97.50%	0
22	Prague	96.98%	0
9	Edinburgh	96.44%	0
8	Dublin	93.57%	0
5	Budapest	85.66%	2
1	Amsterdam	79.85%	5
3	Berlin	big	3
4	Brussels	big	1
6	Cologne	big	0
7	Copenhagen	big	0
11	Geneva	big	3
12	Genoa	big	0
13	Hamburg	big	0
14	Helsinki	big	4
15	Istanbul	big	0
19	Munich	big	0
20	Oslo	big	0
24	Salzburg	big	1
26	Stockholm	big	0
27	Tallinn	big	0

DMU: decision making unit.

Table 7.

DEA results (model 0)—private listings.

	Inefficient DMU	Score	Benchmarking partners and weights
23	Rome	115.55%	1 (0.32) 4 (0.46) 11 (0.19) 14 (0.04)
10	Florence	113.80%	1 (0.12) 8 (0.28) 11 (0.31) 14 (0.29)
28	Vienna	107.40%	3 (0.19) 6 (0.02) 13 (0.07) 14 (0.48) 24 (0.24)
2	Barcelona	106.90%	1 (0.64) 3 (0.01) 8 (0.05) 14 (0.27) 26 (0.02)
12	Genoa	105.30%	4 (0.18) 24 (0.29) 27 (0.53)
20	Oslo	105.08%	1 (0.14) 4 (0.18) 11 (0.06) 14 (0.62)
17	London	104.98%	1 (0.43) 11 (0.32) 26 (0.25)
21	Paris	104.93%	1 (0.13) 3 (0.23) 11 (0.47) 26 (0.17)
22	Prague	103.17%	3 (0.13) 5 (0.10) 14 (0.17) 27 (0.60)
7	Copenhagen	101.65%	1 (0.49) 3 (0.45) 11 (0.03) 26 (0.04)
	Efficient DMU	Score	Benchmark appearance
16	Lisbon	99.29%	0
5	Budapest	96.34%	1
18	Madrid	94.33%	0
25	Seville	93.62%	0
9	Edinburgh	93.51%	0
13	Hamburg	88.96%	1
8	Dublin	86.10%	2
1	Amsterdam	80.21%	7
3	Berlin	big	5
4	Brussels	big	3
6	Cologne	big	1
11	Geneva	big	6
14	Helsinki	big	6
15	Istanbul	big	0
19	Munich	big	0
24	Salzburg	big	2
26	Stockholm	big	4
27	Tallinn	big	2

DMU: decision making unit.

Table 8.

Summary DEA results (model 0)—commercial versus private listings.

	DMU	Score (commercial)	Score (private)
1	Amsterdam	79.85%	80.21%
2	Barcelona	106.04%	106.90%
3	Berlin	big	big
4	Brussels	big	big
5	Budapest	85.66%	96.34%
6	Cologne	big	big
7	Copenhagen	big	101.65%
8	Dublin	93.57%	86.10%
9	Edinburgh	96.44%	93.51%
10	Florence	116.18%	113.80%
11	Geneva	big	big
12	Genoa	big	105.30%
13	Hamburg	big	88.96%
14	Helsinki	big	big
15	Istanbul	big	big
16	Lisbon	100.28%	99.29%
17	London	100.68%	104.98%
18	Madrid	99.11%	94.33%
19	Munich	big	big
20	Oslo	big	105.08%
21	Paris	98.34%	104.93%
22	Prague	96.98%	103.17%
23	Rome	113.47%	115.55%
24	Salzburg	big	big
25	Seville	97.50%	93.62%
26	Stockholm	big	big
27	Tallinn	big	big
28	Vienna	105.54%	107.40%

DEA: data envelopment analysis; DMU: decision making unit.

Overall, the efficiency scores range from 79.85% to 116.18% for commercial listings, and from 80.21% to 115.55% for private listings. Yet, upon inspection of scores from the two separate DEA runs, it can be argued that commercial listings are generally more efficient than the private ones. This has also been corroborated with a rank-order statistical test: the commercial listings are (tightly) significantly more efficient than the private ones in a one-tailed Wilcoxon paired samples rank test (p = 0.049).

Six commercial DMUs are inefficient in contrast to ten private ones. Florence-based commercial listings are the worst performer with 16.18% potential for improvement; the ones of Rome are the second worst with an inefficiency of 13.47%. With regard to private listings, these two DMUs exchange places, with 15.55% (Rome) and 13.80% (Florence) deviations from maximum performance. The most important benchmark for Florence-based commercial listings remains Amsterdam (0.53), yet that is not the case for its private listings, where the private accommodations of Geneva have the highest importance (0.31). The listings of Brussels are confirmed as the most important benchmark for both commercial and private listings of Rome.

On the efficient side, there are 22 commercial and 18 private DMUs that achieve maximum performance, out of which 14 commercial and 10 private are determined to have “big” scores. Of the ones with numerical scores, the listings of Amsterdam head the efficient DMUs in both samples. In addition, the listings of Amsterdam and Helsinki are the most prominent benchmarks with five and four appearances, respectively, in the commercial sample; the same holds with regard to the private sample, with seven (Amsterdam) and six (Helsinki) appearances, now joined by six benchmark appearances of Geneva. What is interesting to observe is the high number of zero benchmark appearances in the commercial sample: 14 out of 22 efficient DMUs are not recognized as best practices. This, however, does not apply to private sample, where two thirds of the efficient DMUs are identified as benchmarks for inefficient DMUs.

In summary of the two separate DEA runs (Table 8), commercial and private samples of 17 cities are regarded as efficient. This means that both commercial and private listings of, for example, Berlin have achieved maximum performance. The listings of five cities (Barcelona, Florence, London, Rome, and Vienna) are also consistent in their (under)performance: inefficient in both samples. Yet, in four of these five cases, a higher level of inefficiency is recorded in the private sample. For instance, private listings of Vienna are 7.40% inefficient, whereas the Viennese commercial ones are 5.54% inefficient. Furthermore, six DMUs are highlighted in Table 8 as they are the most interesting cases to observe due to a very different performance across samples: the listings of Copenhagen, Genoa, Lisbon, Oslo, Paris, and Prague. For all but Lisbon, it can be concluded that their commercial listings outperform their private ones: their commercial listings are efficient, while the opposite holds for their private listings. This is most noticeable in the case of Genoa, which ranges from extremely high efficiency in the commercial to 5.30% inefficiency in the private sample. Thus, all of these examples corroborate the conclusion of more competitive performance of commercial listings. Turning lastly to the aforementioned exception, the Lisbon-based commercial score denotes an inefficiency of 0.28%, whereas the private score secures the efficient classification (99.29%). Given the negligible inefficiency, one could argue that the observed discrepancy between the commercial and private samples in this particular case could be disregarded.

Conclusions

The present study opted to continue the discussion initiated by Zekan et al. (2019) on the competitiveness of the sharing economy, specifically the Airbnb sector of European cities. This was done by following up on the authors’ suggestion about the importance of including hotel-related data as uncontrollable variables into future DEA modeling endeavors within this sector. In order to verify the robustness of results, interactive DEA was again employed on this occasion. In addition, the present study inspected efficiency of not only cumulative but also of commercial and private samples of Airbnb listings of European cities. Thus, by including uncontrollable variables and by inspecting three samples, this benchmarking study delved into the core of the sector and provided in-depth insights about its efficiency. As evident in the literature review on Airbnb, no other study to date has attempted something similar.

So, what are some of the main takeaways? Given that DEA measures relative efficiency, it is of utmost importance to inspect the robustness of findings by applying interactive modeling. In the present study, 12 variables were employed across five models in order to account for different factors in performance estimations for the cumulative sample. This approach revealed that there are number of DMUs that are underperforming, irrespective of the modeling employed. Florence-based listings were oftentimes identified as the most inefficient DMU, with up to 17.82% room for improvement of at least one of the outputs and, as such, were put into the analytical spotlight. Performance estimations give a clear indication about efficient versus inefficient classification, yet they do not suffice if benchmarking is the ultimate aim. The practical value lies in finding ways to move forward, to improve performance. The starting point can be to look into best practices: DEA offers this opportunity by proposing individual rather than general benchmarking partners for every inefficient unit. For instance, in the case of Florence, Amsterdam-based listings were repeatedly proposed, which signals the importance of this particular best practice. Sectoral regulatory bodies and destination management organizations (DMOs) of the two cities would be the obvious candidates for enabling the knowledge exchange (Zekan et al., 2019). Especially the latter could engage more in this exchange given their orchestrating role and communications with all stakeholders (including accommodation providers and regulatory bodies) at their own destinations and connectivity to DMOs of other European cities. This connectivity and knowledge sharing between the cities’ DMOs is very much apparent and regarded as valuable at the European level in particular, an example of which would be European Cities Marketing. Additionally, the optimal output targets have been calculated for Florence-based listings according to the virtual benchmark. Hence, this DMU could work toward these targets in the post-COVID-19 recovery.

The present study corroborates the findings of Önder et al. (2017) and Zekan et al. (2019) with regard to (1) removal of a causal link between “efficient” and “best practice” and (2) absence of a single best practice for all inefficient units. Another important finding is that, generally speaking, the commercial listings are more competitive than the private ones. Differences exist within the listings of single cities, yet even in cases when both samples of a city were inefficient, a higher level of inefficiency was often detected in the private sample. Therefore, the conclusion can be drawn that hosts who own two or more properties are better at resource utilization, which results in the higher efficiency of commercial listings.

This study is not without limitations. Most notably, one cannot generalize these findings to every possible input/output combination, despite the efforts made in this study with regard to interactive modeling. Moreover, the Airbnb listings (cumulative, commercial, and private) of 28 cities were analyzed in order to preserve the number of cities and by doing so, gain insights into the sector’s (not the destination’s!) competitive standing across Europe. Yet, any change in the number of DMUs may lead to different efficiency scores. All analyses covered only one time period. Subject to data availability (e.g., regarding regulatory environment, changed travelers’ preferences, etc.) and inclusion of Airbnb listings of more cities (i.e., more DMUs to further challenge relative efficiency), future research should orientate toward longitudinal analysis for the assessment of competitiveness over time. It would be very interesting to include pre-COVID-19, COVID-19, and post-COVID-19 periods into analyses. Another recommended research endeavor would be a two-stage analysis: regressions of potential environmental variables could be run before DEA with the purpose of detecting whether a potentially uncontrollable variable is influential.

On the whole, the present study has further deepened the competitiveness discussion of the Airbnb sector. Its humble contribution lies in engaging Airbnb listings and hotels into the efficiency discussion, along with looking beyond the cumulative number of listings by dissecting the overall sector into commercial and private listings—something that has not been attempted as of yet, in spite of the ever-growing body of literature on the sharing economy.

Footnotes

Acknowledgments

The authors express their gratitude toward Steve Hood and Melane Rueff from STR, who generously provided the hotel-related data. They also thank Professors Karl Wöber and Ivo Ponocny for their helpful comments, as well as David Leonard for proofreading the manuscript.

Declaration of conflicting interests

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

Funding

The authors thank the consortium consisting of Michaeler & Partner, PKF hotelexperts, and the Vienna Tourist Board that kindly funded the purchase of the Airbnb data.

ORCID ID

Bozana Zekan

Author biographies

Bozana Zekan is an Assistant Professor at the Department of Tourism and Service Management at Modul University Vienna. She holds a Doctor of Social and Economic Sciences (with honors) degree from the Vienna University of Economics and Business, Vienna, Austria, and a Master of Science in Service Management degree from the Rochester Institute of Technology, Rochester, New York. Her research interests are mainly within the field of destination management (e.g., benchmarking, competitiveness, efficiency studies with the application of data envelopment analysis, key performance indicators, etc.).

Ulrich Gunter is an Associate Professor at the Department of Tourism and Service Management at Modul University Vienna. He holds a PhD in Economics and an MA in Latin American Studies from the University of Vienna. Ulrich was a visiting researcher at the University of Surrey, the University of Sao Paulo, and the University of Florida. His research interest is in applied econometrics, with a particular focus on time-series and panel-data analysis. His primary fields of thematic interest are the economics of tourism and sustainability.

References

Abril

(2020) Airbnb’s IPO filing reveals huge COVID impact. Fortune. Available at: https://fortune.com/2020/11/16/airbnb-ipo-initial-public-offering-coronavirus-impact/ (accessed on 12 March 2021).

Adamiak

(2018) Mapping Airbnb supply in European cities. Annals of Tourism Research 71: 67–71.

Adamiak

Szyda

Dubownik

García-Álvarez

(2019) Airbnb offer in Spain-spatial analysis of the pattern and determinants of its distribution. ISPRS International Journal of Geo-Information 8: 155.

Airbnb (2021) About us. Available at: https://news.airbnb.com/about-us/ (accessed 12 March 2021).

Arrow

Chenery

Minhas

, et al. (1961) Capital-labor substitution and economic efficiency. The Review of Economics and Statistics 43: 225–250.

Assaf

Agbola

(2011) Modelling the performance of Australian hotels: a DEA double bootstrap approach. Tourism Economics 17: 73–89.

Assaf

Josiassen

(2016) Frontier analysis: a state-of-the-art review and meta-analysis. Journal of Travel Research 55: 612–627.

Banker

Charnes

Cooper

(1984) Some models for estimating technical and scale inefficiencies in data envelopment analysis. Management Science 30: 1078–1092.

Bauernfeind

Mitsche

(2008) The application of the data envelopment analysis for tourism website evaluation. Information Technology & Tourism 10: 245–257.

10.

Belarmino

Koh

(2020) A critical review of research regarding peer-to-peer accommodations. International Journal of Hospitality Management 84: 102315.

11.

Belk

(2007) Why not share rather than own? The ANNALS of the American Academy of Political and Social Science 611: 126–140.

12.

Belk

(2014) You are what you can access: sharing and collaborative consumption online. Journal of Business Research 67: 1595–1600.

13.

Belleflamme

Lambert

Schwienbacher

(2014) Crowdfunding: tapping the right crowd. Journal of Business Venturing 29: 585–609.

14.

Bolton

Greiner

Ockenfels

(2013) Engineering trust: reciprocity in the production of reputation information. Management Science 59: 265–285.

15.

Cabral

(2015) A dollar for your thoughts: feedback-conditional rebates on eBay. Management Science 61: 2052–2063.

16.

Chamberlin

(1933) The Theory of Monopolistic Competition. Cambridge, MA: Harvard University Press.

17.

Charnes

Cooper

Rhodes

(1978) Measuring the efficiency of decision making units. European Journal of Operational Research 2: 429–444.

18.

Choe

(2007) AirBed and breakfast for connecting ’07. Core77. Available at: http://www.core77.com/posts/7715/airbed-breakfast-for-connecting-07-7715 (accessed on 16 March 2021).

19.

Cohen

Harcourt

(2003) Retrospectives whatever happened to the Cambridge capital theory controversies? Journal of Economic Perspectives 17: 199–214.

20.

Cook

Tone

Zhu

(2014) Data envelopment analysis: prior to choosing a model. Omega 44: 1–4.

21.

Cooper

Seiford

Zhu

(2004) Data envelopment analysis: history, models and interpretations. In: Cooper

Seiford

Zhu

(eds), Handbook on Data Envelopment Analysis. Boston, Dordrecht and London: Kluwer Academic Publishers, pp. 1–39.

22.

Daly

(1997) Georgescu-Roegen versus Solow/Stiglitz. Ecological Economics 22: 261–266.

23.

Dann

Teubner

Weinhardt

(2019) Poster child and guinea pig - insights from a structured literature review on Airbnb. International Journal of Contemporary Hospitality Management 31: 427–473.

24.

Difallah

Catasta

Demartini

, et al. (2015) The dynamics of micro-task crowdsourcing: the case of Amazon mturk. In: Gangemi

Leonardi

Panconesi

(eds), WWW’15: Proceedings of the 24th International Conference on World Wide Web. New York, NY: Association for Computing Machinery, pp. 238–247.

25.

Dixit

Stiglitz

(1977) Monopolistic competition and optimum product diversity. The American Economic Review 6: 297–308.

26.

Dogru

Mody

Line

, et al. (2020) Investigating the whole picture: comparing the effects of Airbnb supply and hotel supply on hotel performance across the United States. Tourism Management 79: 104094.

27.

Dogru

Mody

Suess

(2019) Adding evidence to the debate: quantifying Airbnb's disruptive impact on ten key hotel markets. Tourism Management 72: 27–38.

28.

Dolnicar

Zare

(2020) COVID19 and Airbnb - disrupting the disruptor. Annals of Tourism Research 83: 102961.

29.

Donthu

Hershberger

Osmonbekov

(2005) Benchmarking marketing productivity using data envelopment analysis. Journal of Business Research 58: 1474–1482.

30.

ECM (2020) The European Cities Marketing Benchmarking Report. 16th Official Edition 2019 – 2020. Dijon: ECM.

31.

Edelman

Luca

(2014) Digital discrimination: the case of Airbnb.com. Harvard Business School Working Paper No. 14-054. Cambridge, MA: Harvard University.

32.

Edelman

Luca

Svirsky

(2017) Racial discrimination in the sharing economy: evidence from a field experiment. American Economic Journal: Applied Economics 9: 1–22.

33.

Edwards

(1933) Review: the theory of monopolistic competition by Edward Chamberlain. The American Economic Review 23: 683–685.

34.

Ert

Fleischer

Magen

(2016) Trust and reputation in the sharing economy: the role of personal photos in Airbnb. Tourism Management 55: 62–73.

35.

Eugenio-Martin

Cazorla-Artiles

González-Martel

(2019) On the determinants of Airbnb location and its spatial distribution. Tourism Economics 25: 1224–1244.

36.

Falk

Yang

(2020) Hotels benefit from stricter regulations on short-term rentals in European cities. Tourism Economics. DOI: 10.1177/1354816620918769.

37.

Farrell

(1957) The measurement of productive efficiency. Journal of the Royal Statistical Society. Series A (General) 120: 253–281.

38.

Farronato

Fradkin

(2018). The welfare effects of peer entry in the accommodation market: the case of Airbnb. Cambridge, MA: National Bureau of Economic Research. NBER Working Paper No. 24361.

39.

Grassmuck

(2012) The sharing turn: why we are generally nice and have a good chance to cooperate our way out of the mess we have gotten ourselves into. In: S̈tzl

Stalder

Maier

, et al. (eds), Cultures and Ethics of Sharing / Kulturen und Ethiken des Teilens. Innsbruck: Innsbruck University Press, pp. 17–34.

40.

Green

(2012) Trusting and being trusted in the sharing economy. Forbes. Available at: http://www.forbes.com/sites/trustedadvisor/2012/05/02/trusting-and-being-trusted-in-the-sharing-economy/2/#7b10892b6b5b (accessed on 16 March 2021).

41.

Gunter

(2018) What makes an Airbnb host a Superhost? Empirical evidence from San Francisco and the Bay area. Tourism Management 66: 26–37.

42.

Gunter

Önder

(2018) Determinants of Airbnb demand in Vienna and their implications for the traditional accommodation industry. Tourism Economics 24: 270–293.

43.

Gunter

Önder

Zekan

(2020) Modeling Airbnb demand to New York City while employing spatial panel data at the listing level. Tourism Management 77: 104000.

44.

Guttentag

(2015) Airbnb: disruptive innovation and the rise of an informal tourism accommodation sector. Current Issues in Tourism 18: 1192–1217.

45.

Guttentag

(2019) Progress on Airbnb: a literature review. Journal of Hospitality and Tourism Technology 10: 814–844.

46.

Guttentag

Smith

SLJ

(2017) Assessing Airbnb as a disruptive innovation relative to hotels: substitution and comparative performance expectations. International Journal of Hospitality Management 64: 1–10.

47.

Haucap

Heimeshoff

(2014) Google, Facebook, Amazon, eBay: is the Internet driving competition or market monopolization? International Economics and Economic Policy 11: 49–61.

48.

Hawlitschek

Teubner

Weinhardt

(2016) Trust in the sharing economy. Die Unternehmung 70: 26–44.

49.

Hodak

Krajinović

(2020) Examining recent research in the field of sharing economy in tourism: bibliometric and content analysis. Tourism: An International Interdisciplinary Journal 68: 402–414.

50.

Hossain

(2020) Sharing economy: a comprehensive literature review. International Journal of Hospitality Management 87: 102470.

51.

Kadi

Schneider

Seidl

(2020) Short-term rentals, housing markets and COVID-19: theoretical considerations and empirical evidence from four Austrian cities. Critical Housing Analysis 7: 47–57.

52.

Kim

Liu

Williams

(2021) Competitiveness in the visitor economy: a systematic literature review. Tourism Economics. DOI: 10.1177/13548166211034437.

, et al. (2016) Conceptualization of smart tourism destination competitiveness. Asia Pacific Journal of Information Systems 26: 561–576.

(2020) Peer-to-peer (P2P) accommodation in the sharing economy: a review. Current Issues in Tourism. DOI: 10.1080/13683500.2020.1786505.

, et al. (2021) Exploring the relationship between Airbnb and traditional accommodation for regional variations of tourism markets. Tourism Economics. DOI: 10.1177/1354816621990173.

56.

Messe Berlin (2016) ITB World Travel Trends Report December 2016. Berlin: Messe Berlin.

57.

Mollick

(2014) The dynamics of crowdfunding: an exploratory study. Journal of Business Venturing 29: 1–16.

58.

Molz

(2013) Social networking technologies and the moral economy of alternative tourism: the case of couchsurfing.org. Annals of Tourism Research 43: 210–230.

(2019) Spatial price dependencies between the traditional accommodation sector and the sharing economy. Tourism Economics 25: 1150–1166.

(2017) Towards a sustainable urban tourism development in Europe: the role of benchmarking and tourism management information systems - a partial. Tourism Economics 23: 243–259.

61.

Oskam

(2019) The Future of Airbnb and the Sharing Economy: The Collaborative Consumption of Our Cities. Bristol: Channel View Publications.

62.

Oskam

Boswijk

(2016) Airbnb: the future of networked hospitality businesses. Journal of Tourism Futures 2: 22–42.

63.

Parker

Van Alstyne

(2005) Two-sided network effects: a theory of information product design. Management Science 51: 1494–1504.

, et al. (2013) Understanding the diffusion of public bikesharing systems: evidence from Europe and North America. Journal of Transport Geography 31: 94–103.

, et al. (2016) Who benefits from the sharing economy of Airbnb? In: Bourdeau

Hendler

Nkambou Nkambou

(eds), WWW’16: Proceedings of the 25th International Conference on World Wide Web. Geneva: International World Wide Web Conferences Steering Committee, pp. 1385–1394.

66.

Ranchordás

(2015) Does sharing mean caring? Regulating innovation in the sharing economy. Minnesota Journal of Law, Science and Technology 16: 413–475.

67.

Rasmussen

(2013) Production Economics. Berlin, Heidelberg: Springer.

68.

Reynolds

(2003) Hospitality-productivity assessment using data-envelopment analysis. Cornell Hotel and Restaurant Administration Quarterly 44: 130–137.

69.

Robinson

(1933) The Economics of Imperfect Competition. New York, NY: Macmillan.

70.

Rochet

Tirole

(2006) Two-sided markets: a progress report. The RAND Journal of Economics 37: 645–667.

71.

Sainaghi

(2020) The current state of academic research into peer-to-peer accommodation platforms. International Journal of Hospitality Management 89: 102555.

72.

Sainaghi

Baggio

(2020) Substitution threat between Airbnb and hotels: myth or reality? Annals of Tourism Research 83: 102959.

73.

Sainaghi

Baggio

(2021) Are mom-and-pop and professional hosts actually competing against hotels? International Journal of Contemporary Hospitality Management 33: 808–827.

, et al. (2014) Quantifying the benefits of vehicle pooling with shareability networks. Proceedings of the National Academy of Sciences 111: 13290–13294.

75.

Scheel

(2000) EMS: Efficiency measurement system user’s manual. Available at: http://www.holger-scheel.de/ems/ems.pdf (accessed 08 August 2021).

76.

Schor

(2016) Debating the sharing economy. Journal of Self-Governance and Management Economics 4: 7–22.

77.

Schumpeter

(1942) Capitalism, Socialism and Democracy. New York, NY, London: Harper.

78.

Shaikh

(1974) Laws of production and laws of algebra: the humbug production function. The Review of Economics and Statistics 56: 115–120.

79.

Sickles

Zelenyuk

(2019) Measurement of Productivity and Efficiency. Cambridge, UK: Cambridge University Press.

80.

Skalska

Shcherbiak

(2016) The competitiveness on the tourism market: Airbnb as challenge for small enterprises in the tourism sector Analysis of prices of accommodation services on the examples of cities with strong tourist function. In: Marasco

(ed), 2nd International Workshop on the Sharing Economy. Paris: The Urban Media Lab.

81.

Sonnemaker

(2020) Airbnb is worth more than the 3 largest hotel chains combined after its stock popped 143% on its first day of trading. Business Insider. Available at: https://www.businessinsider.com/airbnb-ipo-valuation-tops-three-hotel-chains-combined-opening-day-2020-12?r=DE&IR=T (accessed on 28 March 2021).

82.

Solow

(1956) A contribution to the theory of economic growth. The Quarterly Journal of Economics 70: 65–94.

83.

Thebault-Spieker

Terveen

Hecht

(2015) Avoiding the south side and the suburbs: the geography of mobile crowdsourcing markets. In: Cosley

Forte

(eds), CSCW '15: Proceedings of the 18th ACM Conference on Computer Supported Cooperative Work & Social Computing. New York, NY: Association for Computing Machinery, pp. 265–275.

84.

Tong

Gunter

(2020) Hedonic pricing and the sharing economy: how profile characteristics affect Airbnb accommodation prices in Barcelona, Madrid, and Seville. Current Issues in Tourism. DOI: 10.1080/13683500.2020.1718619..

85.

Tsai

Zhou

(2011) Managing efficiency in international tourist hotels in Taipei using a DEA model with non-discretionary inputs. Asia Pacific Journal of Tourism Research 16: 417–432.

86.

UNWTO (2021). World tourism barometer. Report 19(1), January. Madrid: UNWTO.

87.

Vavilis

Petković

Zannone

(2014) A reference model for reputation systems. Decision Support Systems 61: 147–154.

88.

Wirtz

KKF

Mody

, et al. (2019) Platforms in the peer-to-peer sharing economy. Journal of Service Management 30: 452–483.

89.

Wöber

(2002) Benchmarking in Tourism and Hospitality Industries: The Selection of Benchmarking Partners. Wallingford: CAB International.

90.

Wöber

Fesenmaier

(2004) A multi-criteria approach to destination benchmarking: a case study of state tourism advertising programs in the United States. Journal of Travel & Tourism Marketing 16: 1–18.

91.

Yang

Mao

(2020) Location advantages of lodging properties: a comparison between hotels and Airbnb units in an urban environment. Annals of Tourism Research 81: 102861.

92.

Yeon

Song

Lee

(2020) Impact of short-term rental regulation on hotel industry: a difference-in-differences approach. Annals of Tourism Research 83: 102939.

93.

Zekan

Önder

Gunter

(2019) Benchmarking of Airbnb listings: how competitive is the sharing economy sector of European cities? Tourism Economics 25: 1029–1046.

94.

Zervas

Proserpio

Byers

(2017) The rise of the sharing economy: estimating the impact of Airbnb on the hotel industry. Journal of Marketing Research 54: 687–705.