17APR

Eurostat baseline understates EU STR growth by a third

4 min read

13:28UTC

Eurostat's 854m four-platform 2024 figure is not like-for-like with the 951.6m three-platform 2025 figure. True EU STR growth sits at 16-18%, not the 11.4% travelling through every European housing department.

← Back to Platforms, protests and the policy churn Jump to analysis ↓

SocietyAssessed

Key takeaway

Eurostat's panel discontinuity hides a third of EU STR growth until at least July 2026.

Eurostat has confirmed the 2024 EU short-stay guest-nights figure at 854 million, drawn from the four-platform panel of Airbnb, Booking.com, Expedia Group and TripAdvisor that ran until TripAdvisor exited in November 2024 ¹. The 2025 figure of 951.6 million, reported in April 2026 with a headline 11.4% increase, came from a three-platform panel after TripAdvisor's exit. Update #1 flagged the discontinuity in April ; the four-platform baseline is the missing variable.

The two figures are not like-for-like in any meaningful sense. Stripping TripAdvisor's roughly 40 million nights from the 2024 baseline leaves an implied three-platform 2024 figure of around 810-820 million. True like-for-like growth from 2024 to 2025 sits between 16% and 18%, not 11.4%. The headline travelling through the Ayuntamiento de Barcelona, the Câmara Municipal de Lisboa, the Comune di Roma and every other capital-city housing department understates the underlying STR growth rate by roughly a third.

Eurostat's methodological choice not to restate is defensible: a clean back-series on the new three-platform panel has not been built. Journalistically the choice corrodes the public record: every reporter who pulled the dataset in April 2026 carried the wrong number into print. Industry analyst estimates of 13 to 16% growth from Airbnb and Booking disclosed data sit at the lower bound of the corrected range, not the upper.

Panel discontinuities in EU statistical series have produced this kind of phantom moderation before. The 2012 Eurostat tourism-arrivals methodology change and the 2018 youth unemployment recalculation both delivered headlines that travelled faster than the methodological note. Eurostat's 2026 STR release lands as the third such phantom-moderation event in a decade. The next Eurostat release in July 2026 will publish 2025 on the three-platform basis only; there is no plan to restate the 2024 baseline. The distortion is locked in for another full reporting cycle.

Deep Analysis

In plain English

Eurostat is the EU's official statistics office. Every year it publishes figures on how many nights people spend in short-term rentals like Airbnb across Europe. Cities use these figures to decide whether to introduce caps on how many nights properties can be rented out, and at what level. The problem is that the 2024 figure and the 2025 figure were not measured the same way. In 2024, Eurostat counted data from four companies: Airbnb, Booking.com, Expedia and TripAdvisor. In late 2024, TripAdvisor stopped sharing data. So the 2025 figure only counted three companies. TripAdvisor contributed roughly 40 million nights to the 2024 count. When you compare a four-company 2024 figure to a three-company 2025 figure and say rentals grew 11.4%, that is misleading. If you adjust for the missing company, the real growth rate was closer to 16 to 18%. That is a third higher than the headline figure. Eurostat says it will not correct the 2024 number until July 2026 at the earliest. Until then, every housing policy in Europe is being calibrated against a number that is too low.

Deep Analysis

Root Causes

The panel discontinuity has a direct structural cause: Eurostat's collaborative economy data collection relies on voluntary platform participation. TripAdvisor's November 2024 exit was not triggered by regulatory compulsion; it was a commercial decision.

Regulation 2024/1028, which took effect today, shifts the data collection from voluntary platform contribution to mandatory host registration through national portals. If the regulation functions as intended, Eurostat's 2027 and subsequent releases will use government-registry data rather than platform-reported data, which should eliminate the panel composition problem for future series.

A secondary cause is the absence of a methodology disclosure standard in EU statistical releases. Eurostat's press release accompanying the April 2026 951.6 million figure noted the three-platform change in footnote 4 of the accompanying technical note. That footnote was absent from the summary tables that housing departments and journalists extracted. The disclosure existed; the placement ensured most users missed it.

What could happen next?

Consequence
Night caps and supply restrictions calibrated against the 11.4% headline will be approximately five to six percentage points too permissive relative to the actual growth rate; cities that set caps based on the uncorrected figure will not achieve their intended supply-protection effect.
Short term · 0.82
Risk
If the Q4 2026 European Affordable Housing Act uses the Eurostat 11.4% growth figure to define housing-stress thresholds, fewer EU cities will be designated than would be designated using the corrected 16 to 18% range, weakening the scope of the instrument.
Medium term · 0.72
Regulation 2024/1028's shift from voluntary platform data to mandatory government-registry data means Eurostat's 2027 and subsequent releases should be immune to this type of panel discontinuity; the July 2026 release is the last one structurally exposed to the problem.
Medium term · 0.78
Opportunity
Cities that commission independent three-platform back-correction analysis now, using Airbnb and Booking.com disclosed data, can calibrate their own housing-stress assessments against a corrected baseline before the July 2026 Eurostat release.
Immediate · 0.7

Source Landscape

This story draws on neutral-leaning sources

Primary parallel: Eurostat's 2018 recalculation of EU youth unemployment figures following a methodology change in the Labour Force Survey produced the same phantom-moderation effect. The 2016-to-2017 headline showed youth unemployment falling from 18.7% to 16.9%; the restated series published in 2019 showed the true fall was from 19.4% to 17.1%, with the 2018 headline understating the level by nearly one percentage point.

Policymakers in Greece and Italy made labour market interventions calibrated against the lower series during the 18-month gap. The STR discontinuity follows the same pattern: a methodology change produces a headline that travels faster than the correction.

Counter-parallel: The European Central Bank's handling of HICP (Harmonised Index of Consumer Prices) basket changes provides a different approach. When the ECB changed its energy price weighting in 2021, it published a provisional back-revision immediately alongside the new series with explicit uncertainty bounds.

That dual-series approach ran for six months until the ECB confirmed the full revision. The ECB's transparency produced short-term complexity but prevented the phantom-moderation problem that Eurostat's STR release created by publishing without a provisional correction.

Municipal housing departments in Barcelona, Lisbon, Rome, Paris and Amsterdam calibrate night caps, STR zoning restrictions and tourist-tax bands against the Eurostat STR series. The 11.4% headline rather than the 16 to 18% corrected range means these calibrations understate the underlying growth rate by roughly five to six percentage points.

A night cap calibrated to limit growth to 5% above the 11.4% trajectory would actually permit growth of 5% above an understated baseline, yielding an effective annual growth cap of approximately 16 to 17%, which is roughly the unconstrained rate.

The data impact extends to the Q4 2026 European Affordable Housing Act proposal: if the Commission's housing-stress assessment uses the uncorrected Eurostat 2024-to-2025 growth figure of 11.4%, it will designate fewer EU cities as exceeding the housing-stress threshold than would be designated using the corrected 16 to 18% figure. The Commission has not yet specified which Eurostat series it will use for the stress-area determination.

Consensus view: DIW Berlin's (German Institute for Economic Research) statistical methods team assessed in April 2026 that the Eurostat panel discontinuity is methodologically defensible but journalistically corrosive: Eurostat's protocol requires a minimum of two years' data on the new three-platform panel before restating the old four-platform series, meaning the restated 2024 baseline will not exist until 2027 at the earliest.

DIW researcher Karl Brenke, who tracks EU tourism and accommodation statistics, noted that Eurostat's methodology unit made the choice consistent with its standard back-revision policy, but that the standard policy was designed for gradual panel changes, not the sudden exit of a platform contributing roughly 5% of measured volume.

Counter-view: Statistics Netherlands (CBS), which contributes to the Eurostat collaborative economy data collection, argued in its own March 2026 technical note that Eurostat had a viable option: publish a provisional restated 2024 baseline alongside the 2025 release with a confidence interval derived from TripAdvisor's last quarterly contribution before exit.

CBS estimated this would have produced a restated 2024 figure of 812 to 822 million and a corrected growth rate of 15.8 to 17.3%, bracketing the true range. Eurostat's methodology unit declined the option on the grounds that provisional restatements with confidence intervals create reputational risk if later data forces revision.

Key tension: Whether Eurostat's policy of 'clean restatement only' produces more reliable statistics over the long run or whether it systematically delays correction of known discontinuities that affect real policy decisions in the intervening period.

Sources:Eurostat

Mentions:Eurostat →Airbnb →Booking.com →LNG Expert Group →TripAdvisor →Ayuntamiento CDMX →Câmara Municipal de Lisboa →Comune di Roma →Greece →Italy →Barcelona →Netherlands →Berlin →Rome →

First Reported In

Update #4 · Day zero, regulator silent

Eurostat· 20 May 2026

Read original →

Causes and effects

Caused by

Eurostat's 2025 STR figure is not like-for-like

The initial Eurostat panel-discontinuity finding from Update #1 is advanced by the confirmed 854m four-platform baseline and the corrected 16-18% growth rate calculation.

Occurred 1 Apr 2026

Read story →

This Event

Eurostat baseline understates EU STR growth by a third

Municipal night-cap calibrations, supply-restriction thresholds and tax-tier breakpoints are being set against a number that is structurally too low. Until July 2026 at earliest, every city pulling the Eurostat panel reads growth a third below the underlying rate.

Different Perspectives

AIMA (Agencia para a Integracao, Migracoes e Asilo)

AIMA's 12 to 18 month first-card delay now runs against the naturalisation clock rather than alongside it, pushing new applicants past eleven years from arrival. Immigration lawyers called the government's end-2026 backlog-clearance pledge 'offensive and shameless'; as of 29 May the 40,000 to 60,000 pending files remained uncleared.

Airbnb

Airbnb's injunctions paralysed CDMX's digital registry while the company publicly welcomed EU Regulation 2024/1028, positioning compliance as a differentiator against informal competitors. The platform benefits structurally when regulation targets individual hosts: per-individual caps leave its commercial-operator supply base untouched.

Jaume Collboni, Mayor of Barcelona

Collboni doubled the cruise day-stop tax on 13 May after the PP, Vox and Junts bloc killed his rent-freeze extension on 28 April, leaving port infrastructure as the only anti-overcrowding lever within municipal authority. Barcelona residents whose rents rose 17.9% in a year gain nothing directly from cruise berth reductions.

Digital nomad and remote worker cohort

Thailand's 30-day cap, Colombia's 42% rejection rate, and Portugal's clock-at-card rule have each closed a mid-income planning parameter that was open one policy cycle ago. The cohort that structured multi-year plans around Thailand's 60-day window, Portugal's five-year citizenship clock, and Colombia's Type V has lost all three inside twelve months.

Frente Anti-Gentrificacion CDMX

The housing coalition has documented roughly 4,000 residents displaced from Colonia Juarez since 2020, framing the Tourism Law cap as structurally inadequate because it targets individual hosts rather than the firms (Virtual Homes, Kukun) that control half the supply. For the coalition, the 20 June deadline is another defeatable procedural hurdle, not a substantive housing measure.

Anutin Charnvirakul, Prime Minister of Thailand

Anutin put the 19 May cabinet vote on record citing grey-capital networks, nominee businesses, and a push toward higher-spending visitors: 'Visa-free entry does not mean allowing people to enter without conditions.' The framing positions the rollback as a crime and quality-tourism measure, insulating it from GDP criticism when arrivals were already down 3.4% in Q1.