Touristification, real-estate rents and capital accumulation through Airbnb. The case of Valencia
Airbnbdigital capitalismtouristificationrent gapValencia
Using a July-2019 InsideAirbnb snapshot, the paper argues that Airbnb in Valencia follows a 'touristifying' rather than 'redistributive' model: listings are mostly commercial, concentrated in tourist districts and among multi-listing hosts, and income is highly concentrated.
Thesis (C1) Airbnb in Valencia has produced a 'touristifying' rather than a 'redistributive' hosting model. adequate
The descriptive core — Valencia's Airbnb supply in 2019 was predominantly entire-home, high-availability, spatially concentrated and controlled disproportionately by multi-listing hosts — is well supported and survives the several arithmetic slips identified (none changes a conclusion). The rent-gap magnitude is an assumption-heavy estimate, and the housing-market effects (supply loss, rent increases, displacement) are inferred from literature and exposure ratios rather than shown for Valencia.
The paper contrasts two ideal-typical hosting models: a 'redistributive' one (Airbnb's advertised peer-to-peer model: residents sharing their underused homes, dispersed across the city) and a 'touristifying' one (professional multi-hosts operating dedicated units in tourist areas). Eight indicators are computed for Valencia from InsideAirbnb (31 July 2019; 7,203 active short-term listings), AirDNA district occupancy rates, Idealista rents and INE housing stock: listing type (p2p vs commercial by >182 days of calendar availability), accommodation type, listing concentration among hosts, district distribution, a district rent gap (STR minus long-term rent), share of the rental stock, guest capacity relative to residents, and income distribution. 76% of listings and ~89% of offered nights are classified as commercial; 58.5% of listings belong to multi-listing hosts; two tourist districts hold 38.7% of listings; converting a home to STR yields an estimated €500-1,200 more per month; entire-home listings equal 10.3% of the rental stock (39.6% in Ciutat Vella); the top 10% of hosts earn 52.8% of revenue. The author concludes Valencia exhibits the touristifying model, which reduces rental supply, raises rents and displaces residents, and calls for regulation.
A city-level diagnostic framework (eight variables) to classify whether Airbnb in a city follows a redistributive or touristifying model, applied to Valencia.
Intro: Airbnb's four advertised benefits. Literature: sharing economy requires idle assets; conversion of homes, rent-gap theory, professionalisation and spatial concentration. Method: eight variables. Results: 4.1-4.8. Conclusion: touristifying model, with claimed effects on rental supply, rents, neighbourhoods and displacement; COVID as regulatory window.
Claims (boxes) coloured by how well the paper supports them; concepts as ellipses. Arrows point from premise to what it supports. Thick black border = thesis; blue border = conclusion/recommendation. Click a node to jump to it. Layout engine: graphviz.
| From | Type | To | Strength | Note |
|---|---|---|---|---|
| K1 | defines | C1 | ||
| K2 | defines | C2 | ||
| K3 | defines | C6 | ||
| C11 | motivates | C1 | moderate | |
| C2 | supports | C1 | strong | |
| C3 | supports | C2 | moderate | |
| C4 | supports | C1 | strong | |
| C5 | supports | C1 | moderate | |
| C6 | explains | C1 | moderate | The rent gap is the proposed mechanism driving professionalisation. |
| C6 | supports | C8 | weak | |
| C7 | supports | C8 | weak | Stock ratio indicates exposure, not effect. |
| C5 | enables | C9 | moderate | |
| C4 | explains | C10 | moderate | |
| C10 | supports | C1 | moderate | |
| C4 | supports | C12 | moderate | |
| C9 | supports | C8 | weak | |
| C1 | motivates | C13 | moderate | |
| C8 | motivates | C13 | moderate |
Airbnb in Valencia has produced a 'touristifying' rather than a 'redistributive' hosting model.
Assessment: Given the paper's own ideal types, the descriptive indicators (commercial share, host concentration, spatial concentration, revenue concentration) consistently point the same way, and the conclusion is robust to the arithmetic slips found. The classification is, however, partly built into the operational thresholds.
About three quarters of listings (75.8%) and close to nine tenths of offered nights are commercial rather than peer-to-peer.
Assessment: Counts are internally consistent (1,745 + 5,458 = 7,203). Nights shares are misreported in one place (11.8%/88.2% instead of 10.87%/89.13%) but the direction is unaffected. The 182-day availability threshold is a proxy whose error is not symmetric: booked nights are excluded from availability, so very busy listings may be classified as p2p (conservative), while hosts who leave calendars open by default may be classified commercial (non-conservative).
Listing ownership is highly vertical: multi-listing hosts are 28.34% of hosts but control 58.54% of listings; 15 hosts control 660 listings.
Assessment: Band counts reconcile exactly to 7,203 listings and 4,167 hosts. Two presentational errors: single-listing hosts are 71.66% (not 71.91%) and hosts with 6+ listings hold 23.4% of listings, not 'almost a third'. Also host accounts are not legal entities: one operator may use several accounts and one account may manage for many owners.
Listings are concentrated in the two tourist districts (38.65%; 46.32% of entire homes), so Airbnb has not significantly spread tourism to other areas; rooms are more dispersed.
Assessment: Shares reconcile with counts. The 'not spread' claim is asserted without a comparison to hotel distribution or to pre-Airbnb tourist geography, but the concentration itself is well documented.
Evidence: E5
Converting a home to a tourist flat yields on average €500-1,200 more per month than long-term letting, and over €1,000/month in tourist districts; this is a conservative estimate.
Assessment: Direction is plausible and consistent with the literature, but the estimate relies on strong assumptions: a uniform 70 m2 dwelling, asking (not contract) rents, list prices, a district-level AirDNA occupancy applied to every listing, and only a 20% cost deduction (no platform fees, taxes, furnishing, licensing, vacancy risk). Some assumptions are conservative, others are not, so the 'conservative' label is not demonstrated. Which listings enter the district mean is not stated.
Entire-home Airbnb listings amount to 10.32% of the city's rental housing stock, and 39.6% in Ciutat Vella.
Assessment: The ratio is a meaningful pressure indicator, but listings are not dwellings (duplicates, tourist-licensed buildings, second homes), the reference year of the INE stock is not given (likely the 2011 census), and the ratio does not show that the listings were withdrawn from the rental stock.
Airbnb income is highly concentrated (top 10% of hosts receive 52.81%), so Airbnb is not a redistributive channel but a means of capital accumulation that widens inequality.
Assessment: The concentration figures are internally consistent, but revenue is modelled (availability x district occupancy x list price), so variation among hosts is mostly driven by number of listings and availability rather than observed bookings. 'Widens social inequality' requires knowing hosts' wealth/income relative to the population, which is not analysed. Also a top-10% share of ~53% is not unusual for any market with heterogeneous supplier size.
Airbnb strongly reduces residential rental supply, raises rents and causes displacement in Valencia.
“reduce la oferta de vivienda en alquiler, ejerce presión sobre los precios de los alquileres, transforma los barrios y produce expulsiones y desplazamientos de la población local”
Assessment: No Valencia rent series, conversion data or displacement data are analysed. The claim is inferred from the stock ratio and from other cities' literature. In section 4.6 it is correctly framed as an expectation ('se espera'), but the conclusion states it as a finding.
Entire-home listings dominate (68.3%) and are listed on average 167 days a year (65.79% > 90 days), which shows they cannot be hosts' habitual residences.
Assessment: The counts are consistent. The inference 'cannot be the habitual residence' is reasonable for listings with high forward availability but availability is not occupancy; an open calendar does not prove non-residence.
In Ciutat Vella, Airbnb capacity equals up to 22.77% of the resident population, transforming the neighbourhood according to tourists' needs.
Assessment: Capacity ratio is explicitly an upper bound (100% occupancy). The neighbourhood-transformation part is not evidenced in the paper.
A more egalitarian listing structure without multi-hosts would likely produce a more equal distribution of income.
Assessment: Nearly tautological given that income is modelled per listing; appropriately hedged ('plausible').
Regulators should use the COVID-19 shock to adopt regulation preventing the touristifying model from re-emerging.
Assessment: Policy recommendation following from the diagnosis; no evaluation of policy options is offered.
No evidence cited in the paper for this claim.
The effects of Airbnb on a city are not intrinsic to the platform but depend on how supply is configured.
Assessment: Framing premise; reasonable and not controversial.
No evidence cited in the paper for this claim.
Quantitative. Cross-sectional descriptive analysis of a single platform snapshot, combined with district-level secondary data; no counterfactual or time series.
Scope: Valencia city, July 2019.
Variables and assumptions are described in prose; no tables of district-level values beyond charts; data year of INE stock, AirDNA period and exact rent-gap formula aggregation (mean over which listings?) not specified. Code not provided.
| ID | Source | Type | Description | Coverage | Loc. |
|---|---|---|---|---|---|
| D1 | InsideAirbnb Valencia scrape (31 July 2019) | scraped platform data | 7,654 listings; inactive (451) and >=30-night (46) listings removed; 7,203 analysed. | City of Valencia, single date. | p. 6 |
| D2 | AirDNA district occupancy rates | scraped platform data | Occupancy rate per district applied to every listing in that district. | Period not specified. | p. 6 |
| D3 | Idealista asking rents | other | Rent per m2 by district, 2019 (excluding Nov-Dec and three peripheral districts). | 2019, most districts. | p. 6 |
| D4 | INE housing stock via Valencia City Council | official statistics | Rental housing stock and population by district. | Reference year not stated. | p. 6 |
| ID | Kind | Description | Loc. | Source | Used by |
|---|---|---|---|---|---|
| E1 | statistic | Listing classification: 1,745 p2p (24.23%) vs 5,458 commercial (75.77%). 1,745 / 5,458 of 7,203 |
p. 9, Gráfico 1 | D1 | C1C2 |
| E2 | statistic | Nights offered: p2p 133,831 vs commercial 1,096,993 per year (text says 11.8% / 88.2%; conclusion says 10.87%). | p. 10, Gráfico 2; p. 20 | D1 | C1C2 |
| E3 | statistic | Entire homes 4,920 (68.3%), private rooms 2,251 (31.25%); entire homes listed on average 167 days; 65.79% >90 days. | pp. 11-12, Gráficos 3-4 | D1 | C3 |
| E4 | statistic | Host concentration: 2,986 single-listing hosts; 1,048 hosts with 2-5 (2,534 listings); 118 with 6-20 (1,023); 15 with >20 (660). Multi-hosts = 28.34% of hosts, 58.54% of listings. | pp. 12-13, Gráfico 5 | D1 | C1C4C12 |
| E5 | statistic | Ciutat Vella 1,551 listings (21.53%), Poblats Marítims 1,233 (17.12%); entire homes in these two districts 46.32% vs rooms 26.12%. | pp. 13-15, Mapa 2, Gráficos 6-7 | D1 | C1C5 |
| E6 | figure | Rent gap per district: €500-1,200/month; >€1,000 in Ciutat Vella, Algirós, Benimaclet, Poblats Marítims; €585-697 in Benicalap, Jesús, Rascanya. | pp. 15-16, Mapa 3, Gráfico 8 | D3 | C6 |
| E7 | statistic | Entire-home listings as share of rental housing stock: 10.32% city; Ciutat Vella 39.6%, Poblats Marítims 27.46%, Poblats del Sud 19.52%; many districts 5-10%. | p. 17, Mapa 4, Gráfico 9 | D4 | C7C8 |
| E8 | statistic | Airbnb bed capacity relative to residents: Ciutat Vella 6,165 places = 22.77%; L'Eixample 6.18%, Extramurs 4.68%, Camins al Grau 3.37%. | p. 18, Mapa 5, Gráfico 10 | D4 | C9 |
| E9 | statistic | Estimated revenue concentration: top 10% of hosts 52.81%, top 20% 67.58%, bottom 80% 32.45%. | p. 19, Gráfico 11 | D2 | C1C10C12 |
| E10 | literature | Literature on conversion of homes to STR raising rents and displacing residents (Barron et al. 2017; Garcia-López et al. 2019; Segú 2018; Cocola-Gant 2016; Wachsmuth & Weisler 2018). | pp. 5-6 | C8 |
| ID | Term | Definition | Origin | Analyst note |
|---|---|---|---|---|
| K1 | Redistributive vs touristifying hosting model p. 3 | Two ideal types: dispersed, resident-hosted, idle-capacity sharing (redistributive) vs professional, concentrated, dedicated-unit hosting in tourist areas (touristifying). | coined | |
| K2 | P2P vs commercial listing pp. 6-7 | Operational: a listing available < 183 days/year in the calendar is p2p; > 182 days is commercial. | operational | Calendar availability is forward-looking and excludes booked nights, so it is a noisy proxy for the share of the year a dwelling is used as a residence. |
| K3 | Rent gap (diferencia de renta) pp. 5, 8-9 | Difference between monthly income from STR and from residential letting of the same dwelling. | adapted Smith; Wachsmuth & Weisler (2018) | Adapted from Smith's ground-rent concept to a gross income differential; ignores conversion, licensing, furnishing and vacancy risks. |
| K4 | Multigestor (multi-listing host) p. 12 | Host controlling two or more listings. | operational |
The conclusion lists reduced rental supply, rent pressure, neighbourhood transformation and displacement as effects of the model, but the paper presents no rent trends, eviction/displacement or conversion data for Valencia; these are inferred from literature on other cities.
Impact on conclusions: These conclusions should be read as hypotheses consistent with the indicators.
The p2p/commercial split and the 'cannot be habitual residence' inference rest on forward calendar availability, which excludes already-booked nights and reflects host calendar settings rather than actual use. Bias direction is ambiguous, so the 'most lenient possible' characterisation is not established. No sensitivity analysis (e.g. 90- or 120-day thresholds, or reviews-based activity).
Impact on conclusions: Magnitudes uncertain; direction likely robust given the size of the gap.
Uniform 70 m2 size, asking rents from Idealista, list prices, district-average occupancy applied uniformly, 20% cost deduction, no platform fees/taxes/licence/furnishing costs. The author argues all choices are conservative, but list prices and omitted costs bias upward. No sensitivity ranges are reported and the averaging unit is unclear.
Impact on conclusions: Existence of a positive gap is credible; the €500-1,200 magnitude is uncertain.
Affects: C6
Entire-home listings are divided by the (undated) INE rental stock. Listings may come from owner-occupied, second or previously vacant homes, may be duplicated, or be in purpose-built tourist apartments; thus the ratio is not a measure of units removed from long-term rental.
Impact on conclusions: Weakens the step from exposure to impact.
(a) 7,654 − 451 − 46 = 7,157, not the 7,203 analysed (7,203 = 7,654 − 451 only; the 46 long-stay listings appear not to have been removed, or overlap with inactive ones). (b) p2p nights are reported as 11.8% / commercial 88.2% on p. 10, but the reported counts give 10.87% / 89.13% (the conclusion uses 10.87%). (c) p2p listings given as 24.19% on p. 11 vs 24.23% elsewhere. (d) Single-listing hosts 71.91% vs 71.66% computed from counts. (e) Hosts with 6+ listings hold 23.4% of listings, described as 'casi un tercio'. (f) >20-listing hosts: 9.16% not 9.14%.
Impact on conclusions: None on direction of conclusions; indicates limited proofreading.
Concentration is measured per host account; management companies may use several accounts and single accounts may represent many owners. The bias likely understates concentration among operators but overstates concentration of ownership.
Impact on conclusions: Qualifies interpretation of 'who' controls supply.
Affects: C4
Revenue is computed from availability, district occupancy and list price, so cross-host dispersion is largely mechanical (number of listings x availability). The step from revenue concentration to 'widening social inequality' needs information on who the hosts are.
Impact on conclusions: Concentration finding stands; inequality interpretation is an extrapolation.
Affects: C10
Scope internal = numbers printed in the document; cross paper = other corpus documents; external / replication = outside sources or re-analysis of original data (web mode only).
| ID | Status | Scope | What was checked | Computation | Result | Loc. | Claims |
|---|---|---|---|---|---|---|---|
| V1 | fail | internal | Sample size after exclusions. | 7,654 − 451 − 46 | = 7,157, but 7,203 is used throughout (7,203 = 7,654 − 451). | p. 6 | C2 |
| V2 | pass | internal | P2P vs commercial listing shares. | 1,745/7,203; 5,458/7,203 | 24.23% / 75.77% — matches. | p. 9 | C2 |
| V3 | minor discrepancy | internal | Nights shares from reported counts. | 133,831 / (133,831 + 1,096,993) | 10.87% p2p / 89.13% commercial; text on p. 10 reports 11.8% / 88.2% (conclusion p. 20 correctly uses 10.87%). | p. 10 | C2 |
| V4 | pass | internal | Accommodation types reconcile to total. | 4,920 + 2,251 + 32 shared rooms | = 7,203; 68.30% / 31.25% / 0.44% — matches. | p. 11 | C3 |
| V5 | minor discrepancy | internal | Host-band shares. | Hosts 2,986+1,048+118+15 = 4,167; shares 71.66 / 25.15 / 2.83 / 0.36% | Text gives 71.91% for single-listing hosts (bands then sum to 100.25%); the complement 28.34% is correct. | pp. 12-13 | C4 |
| V6 | pass | internal | Listing-band shares and averages. | 2,986+2,534+1,023+660 = 7,203; 41.45 / 35.18 / 14.20 / 9.16%; means 2.42, 8.67, 44.0 | All match except 9.16% vs reported 9.14%. | pp. 12-13 | C4 |
| V7 | fail | internal | 'Hosts with six or more listings hold almost a third of listings'. | (1,023 + 660) / 7,203 | 23.4% — closer to a quarter than a third. | p. 12 | C4 |
| V8 | pass | internal | District shares. | 1,551/7,203; 1,233/7,203 | 21.53% and 17.12% (sum 38.65%) — matches. | p. 13 | C5 |
| V9 | pass | internal | Rent-gap annualisation. | €585 x 12 | €7,020 > €7,000 — consistent. | p. 15 | C6 |
| V10 | pass | internal | Revenue shares complement. | 67.58% + 32.45% | 100.03% — rounding. | p. 19 | C10 |
| V11 | not checkable | internal | Implied Ciutat Vella population. | 6,165 / 0.2277 | ≈ 27,075 residents — plausible for the district; not otherwise checkable. | p. 18 | C9 |
Evaluation frame: A descriptive quantitative case study using scraped platform data. It should be judged on the validity of its operational definitions, internal arithmetic consistency, appropriate framing of modelled estimates, and whether causal claims about housing effects are backed by data on those effects.
| Dimension | Score | Rationale |
|---|---|---|
| conceptual clarity | 4 | The two ideal-type models and the eight indicators are clearly set out and mapped onto the research questions. |
| logical coherence | 3 | Indicators consistently support the touristifying diagnosis; the leap to realised rent and displacement effects is not bridged. |
| evidential support | 3 | Strong descriptive evidence on supply structure; modelled rent-gap and revenue; no outcome data on rents/displacement. |
| methodological rigor | 3 | Reasonable standard for InsideAirbnb studies, but single snapshot, proxy thresholds without sensitivity analysis, and uniform occupancy assumptions. |
| transparency | 3 | Sources and assumptions described; several parameters (INE year, AirDNA period, averaging) missing; several arithmetic slips. |
| proportionality | 2 | Conclusion states displacement and rent effects as established findings. |
Gil (2023) states that during the pandemic 'STRs did not return to the residential market (Gil et al., 2023)' and uses this to argue for housing-policy (not tourism-policy) responses. Gil, Vidal & Yrigoy (2026) cite the same source for STR units being 'temporarily redirected to residential or mid-term segments'. Sequera et al. (2022) document central Madrid long-term listings rising from 1,300 to 4,600 and back to 1,900, and Gil (2021) reported hosts 'directly ending the business and renting in the residential market'. The evidence supports a temporary switch, largely under seasonal contracts; the 2023 paper's unqualified statement understates it.
Implication: The policy argument in Gil (2023) should be qualified: switching occurred but was temporary and often outside long-term residential contracts.
Papers: Gil (2023)Gil et al. (2026d)Sequera et al. (2022)this paper
Gil (2021) and Gil (2023) analyse the identical 31 July 2019 InsideAirbnb Valencia dataset. Errors propagate (7,654 − 451 − 46 ≠ 7,203; single-host share 71.91% instead of 71.66%). The rent-gap formula changes (availability-based vs occupancy with 15% commission) giving €500-1,200 vs €497-958 per month, without acknowledgement; the 'not habitual residence' threshold shifts from 90/182 days to 60 days.
Implication: Readers cannot tell which estimate is preferred; magnitudes are assumption-driven.
Papers: this paperGil (2023)
Availability >182 days defines commercial listings in Gil (2021) and Gil & Sequera (2020); Gil (2021) also uses >90 days to argue non-residence; Gil (2023) uses >60 days to estimate homes 'extracted from the rental market' and displacement.
Implication: Lower thresholds mechanically raise displacement estimates; sensitivity analysis is never reported.
Papers: this paperGil & Sequera (2020)Gil (2023)
Analyst: Claude (Anthropic), LLM-generated audit · 2026-11-08 · access: document only · text: text/gil_2021_valencia_airbnb.txt · schema v1.1
No external sources were consulted (offline run). Figures are assumed to be correctly reported; claims[].external, citation checks and external/replication checks were not performed.
Charts were not machine-readable; only numbers printed in the text were checked.