Overcoming the pitfalls of economies-of-scale in shared accommodation: How can effective multi-listing management enhance the sustainability of homestay businesses?

The study addresses whether and how economies-of-scale from multi-listing management influence the survival of homestay businesses, distinguishing between professionalized (standardized, hotel-like, multi-unit) and personalized (unique, host-centric) offerings. In hospitality, where firms face high exit risk, leveraging scale to reduce costs and improve profits is critical, yet shared accommodations differ from manufacturing due to heterogeneous products and host-driven service interactions. Prior work often treats homestays as homogeneous and overlooks differences between professionalization and personalization, as well as time dynamics in survival. The paper posits that professional operations may initially benefit from scale through expertise, standardization, and lower marginal costs, whereas personalized operations may confer long-term social value (authenticity, interaction, belonging) that supports survival. It further proposes that host-guest communication and product quality signals (hard: price; soft: Superhost) may moderate scale effects. The research aims to determine which business type benefits more from economies-of-scale over time, how scale operates within each type, and how communication and quality signals optimize survival under multi-listing management.

Survival analysis in hospitality has examined financial (profitability, indebtedness) and non-financial drivers (location, firm attributes, infrastructure), with increasing interest in sharing accommodations but limited differentiation by business type. Professional homestays are typically defined by managing more than 10 listings and adopting standardized, hotel-like operations; personalized homestays emphasize uniqueness and social value. Literature suggests professional multi-listing hosts gain in revenue/pricing and operational efficiency, while personalized hosts provide social value that can enhance satisfaction and loyalty. However, heterogeneity in survival mechanisms and temporal dynamics has been underexplored. Hypotheses: H1a: Professional properties with economies-of-scale have higher short-term survival than personalized. H1b: Personalized properties with greater social value have higher long-term survival than professional. H2a: Within personalized, more economies-of-scale (more listings) reduces survival. H2b: Within professional, more economies-of-scale increases survival. H3a/H3b: High-quality host-guest communication mitigates negative scale effects in personalized and strengthens positive scale effects in professional. H4a/H4b: Higher price (hard quality signal) mitigates negative scale effects in personalized and enhances positive scale effects in professional. H4c/H4d: Prior Superhost (soft quality signal) mitigates negative scale effects in personalized and enhances positive scale effects in professional.

Data: Airbnb listings from New York City, Los Angeles, Chicago, and Austin, Feb 2019–Jun 2020, covering pre-pandemic, onset, and initial control phases. Listings delisted during observation are coded as exits (failure). To address left truncation, only listings that entered during the observation window were retained, yielding 551,605 observations for 93,209 properties. Average survival time was 5.2 months; exit rates: 52.8% (cross-sectional survival sample) and 8.9% (panel survival sample). After removing missing data, final sample: 458,396 observations (379,090 personalized; 79,306 professional). Variables: Failure (1 exit, 0 survive), St (duration), ProBusi (professional business), Listings (host’s number of listings), Communication (response rate), Price (nightly rate; lagged), Superhost (prior Superhost status; lagged), room type dummies, Accommodates, Amenities, Beds, HouseRules, Distance to city center, Competition in ZIP, ReviewNum, ReviewScore, Cancellation policy strictness, LocalHosts indicator, Covid dummy (post Jan 2020). High-variance variables (Listings, Price, Competition, Distance, ReviewNum, ReviewScore) are log-transformed. Business type operationalization: professional if host manages >10 listings (main specification), with robustness using alternative definitions (availability >182 days; any multi-listing). Empirical strategy: Panel survival analysis with random effects to account for repeated observations and unobserved heterogeneity; both Cox proportional hazards and exponential PH models used. Proportional hazards assumption tested via Schoenfeld residuals (satisfied). Models: (1) Baseline effect of professional business on hazard; (2) Effect of economies-of-scale within type via LogListings; (3) Moderation by Communication, LogPrice, and Superhost via interaction terms with LogListings. Robustness: Accelerated failure time (AFT) models; alternative professionalization definitions; non-parametric Kaplan–Meier curves and log-rank tests.

Non-parametric: Kaplan–Meier survival curves intersect; within first 3 months, professionalized properties exhibit higher survival; beyond 3 months, personalized properties outperform. Log-rank tests p<0.001 for differences short- and long-term. Model 1 (type effect): Within 3 months, ProBusi reduces hazard (Exponential β=-0.047, HR=0.955; Cox β=-0.036, HR=0.964; p<0.01/0.05), supporting H1a. After 3 months, ProBusi increases hazard (Exponential β=0.169, HR=1.184; Cox β=0.158, HR=1.171; p<0.01), supporting H1b (personalized outperform long-term). Model 2 (scale within type): Personalized: more listings increase hazard (e.g., HR≈1.23–2.75 across specs; p<0.01), supporting H2a. Professional: more listings decrease hazard (e.g., HR≈0.84–0.85; p<0.01), supporting H2b. Model 3 (moderators): Personalized: LogListings×Communication negative (HR≈0.75; p<0.05–0.01), supporting H3a; LogListings×LogPrice negative (HR≈0.80; p<0.01), supporting H4a; LogListings×Superhost not significant (fails H4c). Professional: LogListings×Communication negative (HR≈0.49–0.52; p<0.01), supporting H3b; LogListings×LogPrice not significant (fails H4b); LogListings×Superhost positive (HR≈1.61; p<0.01), opposite of H4d, indicating prior Superhost dampens the scale benefit. Additional insights: COVID-19 period associated with higher survival probabilities versus pre-COVID in Kaplan–Meier plots. Quadratic checks suggest for personalized hosts, survival peaks at very small scales (1–2 listings), indicating an early turning point; for professional hosts, scale–survival relation is linear and beneficial. Robustness: AFT models and alternative professionalization definitions yield consistent signs and significance with main results.

Findings clarify how economies-of-scale affect survival heterogeneously across homestay business types and over time. Professional operations gain short-term survival advantages via expertise, standardized processes, and reduced marginal costs, but these advantages wane as the social value of personalization (authenticity, interaction, belonging) becomes more salient, driving superior long-term survival for personalized hosts. Within-type analyses show a scale trap for personalized hosts: expanding listings strains scarce resources (time, attention), eroding social value delivery and increasing exit risk; conversely, professional hosts realize genuine scale economies that reduce hazard with more listings. Communication is a critical lever: strong host-guest interaction both offsets scale-induced personalization losses and amplifies professional scale benefits. Quality signaling operates asymmetrically: hard signals (higher prior prices) alleviate scale’s harms for personalized hosts but do not enhance professional scale benefits; soft signals (prior Superhost) do not help personalized hosts and can undermine professional scale advantages, consistent with expectancy violation when elevated expectations are unmet. The study identifies four pitfalls in pursuing scale: overestimating long-run professionalization benefits; overoptimism about scale in personalized settings; chasing scale via many low-quality personalized listings; and relying on Superhost certification to extend listing life, especially for professional hosts. These insights refine economies-of-scale theory for heterogeneous, interaction-intensive services and inform survival strategies in platform-based lodging.

The study advances understanding of economies-of-scale in shared accommodations by separating professionalization from personalization and examining temporal dynamics. Economies-of-scale confer short-term survival benefits to professional homestays, whereas personalization’s social value enables superior long-term survival. Within types, scale harms personalized survival but benefits professional survival. Effective host-guest communication consistently improves outcomes, mitigating harms for personalized and reinforcing gains for professional hosts. Hard quality signals (higher prior prices) help personalized hosts under scale, while prior Superhost status can erode the scale advantage for professional hosts. Contributions include: extending economies-of-scale theory beyond manufacturing to heterogeneous service platforms; enriching hospitality survival analysis with business-type heterogeneity; and integrating signaling with expectancy violation to explain asymmetric moderation by quality cues. Managerial implications urge personalized hosts to prioritize social value and limit scale (1–2 listings) or invest heavily in communication and high-quality assets if scaling; professional hosts should exploit scale while integrating social value, emphasize communication, avoid overreliance on soft badges, and periodically refresh offerings. Platform policies should reconsider Superhost criteria to reflect long-term performance and social value. Future research should employ broader geographies and time horizons, alternative professionalization measures, and richer behavioral measures of communication and quality.

Data only capture online listing presence/absence on Airbnb; exits may reflect migration to other platforms rather than true failure. No universally accepted definition of professional homestay businesses; although three operationalizations were tested (>10 listings; >182 days availability; multi-listing), generalizability across alternative definitions warrants further validation.