Was there a rush to register tourist dwellings before Spain's community-approval rule of 3 April 2025? Evidence from the Comunitat Valenciana's open registry

Abstract

The rule. Since 3 April 2025, a flat in a building under Spain's Ley de Propiedad Horizontal (LPH) may be let to tourists only with the owners' association's express approval, voted by three fifths of the owners and of the shares. An owner who was already letting under the regional tourism rules may carry on. Filing the regional declaration before 3 April was therefore a way to start without a vote. Whether it protects a filer who was not yet letting is a matter of interpretation.

The data. We counted the 90,091 dwellings in the Comunitat Valenciana's open tourist registry by registration date. From the cadastral part of each address, read before the address was dropped, we split them into:

• units in a divided building or complex, which the rule reaches;
• properties that take up a whole cadastral parcel, mostly detached houses, which it does not.

There was a registration rush, concentrated in two municipalities.

• Size. In the 14 days to 2 April 2025 there were 1,093 registrations against 456 expected (2.40 times). No spaced placebo window in 2023–2026 came close (p = 0.013 over 78 windows). Allowing for late March's usual rise, the excess is about 440.
• Where. 55% of the excess is in two municipalities:
  • València city (+200), which was preparing its own ban on tourist flats in residential buildings;
  • Torrevieja (+151), whose registrations stayed high after the deadline.
• Flats against houses. Within municipalities, flats rose more than houses (contrast 1.73, over 35 municipalities). The difference is not statistically clear (placebo p = 0.089; calibrated 95% interval 0.92–3.22).
• Cause. The rush cannot be attributed to the LPH rule alone.

Registrations then fell, mostly for flats and houses alike.

• Size. From September 2025 to August 2026 there were 6,253 registrations against 13,154 in September 2023–August 2024 (ratio 0.48, CI 0.39–0.57). Against 2022, before a boom, the ratio is 0.88, or 0.72 corrected for cancellations.
• Timing. The fall starts in May 2025, equal for both kinds.
• The flats-specific fall. From October 2025 flats fell further than houses. Paired by calendar month, flats' odds against houses were 0.83 (95% CI 0.71–0.97) of their 2022–23 level. A modest flats-specific fall may exist, but it cannot be attributed to the rule.
• The Generalitat's statement. It called the first year after August 2024 «casi el doble que en el periodo anterior». The file shows 0.84.

What the date means, and what is missing.

• The date. A strong test shows it is set before the registry number, and the filing date is the most plausible reading.
• Survivorship. 91.3% of pre-period registrations survive in today's file, about equally for buildings and chalets.
• The national registry. It did not take effect on 3 April.
• Andalucía and País Vasco. Both peak before the deadline, but in date fields that behave like processing dates.
• Catalonia. The LPH does not apply there, and it publishes no registration date.

1. Introduction

Rules with a start date and a grandfather clause invite a rush. If starting before the date needs no one's permission and starting after it does, you move early. Studies of taxes and benefits measure this as "bunching" at a notch. Studies of short-term rental regulation have mostly measured what happens after a rule, not the rush before it.

Spain offered such a case. Ley Orgánica 1/2025 was published on 3 January 2025 and took effect three months later, on 3 April. From that day, a flat in a building under the LPH needs the express approval of the owners' association before it can start operating as a tourist dwelling. Owners who were already operating under the regional tourism rules keep their right.

The regional tourist registries record when each dwelling was registered. One of them, the Comunitat Valenciana's, is published daily as open data with a registration date for every dwelling. A scouting pilot found 287 registrations on 2 April 2025, the day before the rule, and a slower pace afterwards. This study audits that pilot from scratch and resolves what could undo it:

• what the registry's date means;
• that cancelled dwellings vanish from the file;
• whether other regions show the same pattern;
• which other measures changed around April 2025.

It adds a control group inside the same registry. The rule reaches only dwellings in a building under horizontal property, while the regional and national measures of the period apply to detached houses as well. Municipal measures do not, which is why the comparison is made within municipalities.

2. Legal frame (quoted as in force on 4 October 2026)

All texts were read in the Boletín Oficial del Estado (BOE) consolidated versions through its open-data API, or in the Diari Oficial de la Generalitat Valenciana (DOGV), on 4 October 2026.

2.1 The community-approval rule

Ley Orgánica 1/2025, de 2 de enero (BOE-A-2025-76), was published on 3 January 2025. Its final provision 38.1 reads: «La presente ley orgánica entrará en vigor a los tres meses de su publicación en el Boletín oficial del Estado.» The BOE's note to the LPH says the new paragraph «entra en vigor el 3 de abril de 2025, según determina su disposición final 38.1».

Final provision 4, paragraph one, adds art. 7.3 to the LPH:

«3. El propietario de cada vivienda que quiera realizar el ejercicio de la actividad a que se refiere la letra e) del artículo 5 de la Ley 29/1994, de 24 de noviembre, de Arrendamientos Urbanos, en los términos establecidos en la normativa sectorial turística, deberá obtener previamente la aprobación expresa de la comunidad de propietarios, en los términos establecidos en el apartado 12 del artículo diecisiete de esta Ley.»

Paragraph two rewrites art. 17.12:

«12. El acuerdo expreso por el que se apruebe, limite, condicione o prohíba el ejercicio de la actividad […] requerirá el voto favorable de las tres quintas partes del total de los propietarios que, a su vez, representen las tres quintas partes de las cuotas de participación. […] Estos acuerdos no tendrán efectos retroactivos.»

Before 3 April 2025, art. 17.12 (as worded by Real Decreto-ley 7/2019) let a community limit or condition the activity by the same majority. It required no prior approval.

2.2 Who is protected

The protection is not an additional provision of LO 1/2025 itself. Final provision 4, paragraph three, adds a second additional provision to the LPH:

«Aquel propietario de una vivienda que esté ejerciendo la actividad a que se refiere la letra e) del artículo 5 de la Ley 29/1994 […], con anterioridad a la entrada en vigor de la Ley Orgánica de medidas en materia de eficiencia del Servicio Público de Justicia, que se haya acogido previamente a la normativa sectorial turística, podrá seguir ejerciendo la actividad con las condiciones y plazos establecidos en la misma.»

The provision has two limbs. The owner must be «ejerciendo la actividad» before 3 April 2025, and must have «acogido previamente a la normativa sectorial turística», that is, have filed the regional declaration; the national registry plays no part. An owner who filed on 2 April but had not yet started letting meets only the second limb. Ley 15/2018 art. 53.3 also gives two months to start effectively. Whether such last-day filers are protected is therefore an interpretation, not settled law.

Since Real Decreto-ley 7/2019, agreements under art. 17.12 «no tendrán efectos retroactivos». Owners already letting were therefore already shielded from later prohibitions. What 3 April changed was the default: a new entrant now needs a prior approving vote.

As of 4 October 2026, no amendment, annulment or constitutional challenge touches art. 7.3, art. 17.12 or this provision. The challenges recorded against LO 1/2025 concern other provisions.

2.3 The national registry does not share the date

The national short-term rental registry was set up in the same months, so its dates had to be checked. It did not start on 3 April. Real Decreto 1312/2024 (BOE-A-2024-26931), final provision 4:

«El presente real decreto entrará en vigor el día 2 de enero de 2025, desplegando efectos sus disposiciones el 1 de julio de 2025 […].»

The Ministry's note of 2 June 2025 says applications «se pueden realizar desde el 2 de enero de 2025 […] y hasta el 1 de julio». No provision or official act we read ties the national registry to 3 April 2025.

It matters later, in another way. Art. 10.1 required the property registrar to check, before issuing a number, «la no existencia de elementos obstativos […] de acuerdo con la normativa aplicable en cada caso y los posibles acuerdos de la comunidad de vecinos conforme a la Ley 49/1960». That check applied to applications from 2 January 2025. The number became compulsory for platforms' listings from 1 July 2025, and most applications came in June. It is a second channel through which the community rule could bite, but neither its timing nor this study tests it. The Supreme Court annulled arts. 5, 8 to 10 and other provisions in judgments of 19 May, 21 May and 1 June 2026. Under LJCA art. 72.2 an annulment has general effects from its publication, which came in the BOE on 8 June, 26 June and 18 July 2026.

2.4 What "fecha de alta" means in the Valencian procedure

A tourist dwelling in the Comunitat Valenciana starts with a declaración responsable. Ley 15/2018 art. 53.2 says owners declare «el inicio […] de su actividad» and «posteriormente, el órgano competente inscribirá de oficio a las empresas en el Registro de Turismo». Decreto 10/2021 art. 23 bis.2, as amended by Decreto-ley 9/2024, says the declaration «habilita para el ejercicio de la actividad durante un plazo de cinco años desde su presentación». Ley 39/2015 art. 69.3 says declarations allow the activity «desde el día de su presentación».

Neither text sets a deadline for the administration to inscribe. No norm defines the open-data field fecha_alta. The dataset describes it as «Fecha de alta en el registro», and signatura as «Número de inscripción en el registro». Taken at face value, that is the inscription date. Whether it records the filing or the inscription must be read from the data (§5.1).

2.5 Other dated measures

Only the LPH rule falls on 3 April 2025. But the city of València was preparing, from 31 January 2025, a ban on tourist flats in buildings with residents, and that gave the city's owners a second reason to file early. We could not find a municipal act of València dated 2 or 3 April.

2.6 Catalonia

Horizontal property in Catalonia is governed by the Codi civil de Catalunya, book five (Llei 5/2006, arts. 553-1 and following). Catalan civil law applies «con preferencia a cualesquiera otras» (CCCat art. 111-5). The Catalan code lets statutes limit activities in private units (art. 553-11.2.e), but it has no counterpart to LPH art. 7.3. The 3 April rule therefore did not reach Catalonia. LO 1/2025 itself has no territorial clause for final provision 4: the exclusion follows from art. 149.1.8.ª of the Constitution and from the Catalan code.

3. Prior work

We searched on 4 October 2026 and logged every query in data/prior_work_search.csv (68 queries).

• arXiv (API, 8 queries): nothing on registration ahead of a rule. arXiv 2211.16617, on breaches of short-term-let rules in Ireland, is background only.
• Crossref (21 queries in English and Spanish).
• GitHub (gh, 11 queries).
• Web and official sources (28 queries).

We found no published daily or weekly count of Valencian registrations around 3 April 2025, nor of any other region's. That is narrower than saying none exists: our web search ran out of budget, and local outlets, for instance in Andalucía, may have reported their own records.

• The only public analysis of the Valencian registry on GitHub counts registrations and cancellations by year.
• The Generalitat published totals only (§5.6).
• Terra Meridiana, an English-language outlet read on the Costa Blanca, wrote on 21 April 2025 that «the last-minute rush to register prior to April 3 caused the system to crash». It tied that rush to the national registry («As of April 3, landlords must register their properties with the Registro Único»). That was wrong: the national registry ran from January to July
  1. It gave no figure. Its readers, however, were told that 3 April was a registration deadline, which is a channel for an early filing rush other than the LPH's approval requirement.

Close work on other cases:

• Portugal (Barbosa, Pinto Machado and Terto, 2026). Uses the full national register to estimate the entry effect of the 2023 suspension of new registrations.
• US cities (Guha, Meggiorin and Belo, 2026). Find Airbnb hosts adjusting supply between approval and enforcement of a rule.
• Permits as options (French and Vreugdenhil, 2026). In Colorado's oil and gas reforms, «permits surged before regulation but drilling did not», because «permits lock in the regulatory regime at approval, allowing firms to acquire the option to invest under current rules».
  • This is the closest precedent: a rush to secure the old regime without the activity.
  • The Spanish rule protects only owners who were «ejerciendo la actividad», so a pure option filing may not work here as it did there.
  • The registry cannot tell whether last-day filers were letting.
• Canarias (press). The vacation-rental registry gained 35.6 dwellings a day while the regional law was being drafted (elDiario.es, 3 April 2025).
• The national registry's own deadline. It drew «más de 125.000 solicitudes» in its last month, out of 199,686 (EFE, 29 June 2025).

Our method borrows the vocabulary of bunching at notches (Kleven 2016; Kleven and Waseem 2013) and of grandfathering (Shavell 2007). The regulation-effects literature measures listings after rules, not registrations before them:

• Barcelona (Garcia-López et al. 2020);
• Chicago (Jin, Wagman and Zhong 2024);
• Los Angeles and New Orleans.

What we add is narrow: a daily count of a tourist-registry rush, with a control inside the registry. The one Spanish study we found that uses a regional registry over time covers Mallorca in 2016–2022 (Pons Esteva 2025). The legal doctrine on art. 17.12 has no counts.

4. Data and method

4.1 The registry and its copies

The Generalitat Valenciana publishes its tourist-dwelling registry daily on dadesobertes.gva.es (dataset tur-gestur-vt, Creative Commons Attribution): one row per registered dwelling, with its registration number, municipality and fecha_alta. The file holds live entries only. We used four copies:

• 4 October 2026 (ours): 90,091 dwellings. This is the main copy.
• 14 September 2026, the Internet Archive's only capture: 90,043 dwellings.
• The GVA's "historical" list (dades-turisme-habitatges-comunitat-valenciana-2025): 182,893 entries, 81,671 of them cancelled, each with its cancellation date and a building type.
  • We infer that it was extracted on 10 January 2025: its latest registration and cancellation dates are both of that day, while the file name says 19600101.
  • It keeps essentially every number ever issued, including cancelled ones.
  • A "last period" list of 24 January 2025 accompanies it.
• Study S11's copy of 3 October 2026 was imported and checked by SHA-256, but is not used.

The source files carry names, addresses and cadastral references. These were dropped while reading, after the derivations of §4.2.

We analyse dwellings dated from January 2023 to August 2026: 38,933 registrations. September 2026 is left out because it is still being filled in.

4.2 Which dwellings the rule reaches

The address in the registry ends with the cadastral unit, written ES:<stair> PL:<floor> PT:<door>. A property that is not divided into units reads ES:T PL:OD PT:OS. We call it whole_parcel; any other floor or door makes a unit in_building. Three checks:

• 98.5% of whole-parcel dwellings have cadastral unit number 0001, and 4.6% share their parcel with another registered dwelling.
• For in-building dwellings the figures are 3.7% and 84.8%.
• Against the GVA's own building type, 97% of dwellings "in a block or building" are in-building, and 78% of "chalets or villas" are whole-parcel.

Bungalows and terraced houses are mostly in-building, because they are divided units of a complex. Complexes fall under the LPH too.

in_building is the exposed group; whole_parcel is the control. 1,211 dwellings have no unit part. We call them unknown and, as a sensitivity, count them as exposed.

4.3 Measures

The plan (METHOD.md §1) was frozen at 12:34 UTC on 4 October 2026, before any measure was computed. By then the pilot and the monthly counts by kind were already known.

• Peak. The count on 2 April 2025 (D) against the working days of 7 January–28 February 2025.
• Run-up excess. Observed registrations in the 14 days to D, against an expected count: the baseline mean per day type (working or non-working) from D−87 to D−34, applied to each day of the window.
• Placebo. The same statistic for every other working day from April 2023 to August 2026 taken as the window end, leaving out March–May 2025. That gives 791 windows.
• Contrast. The ratio of the exposed group's run-up ratio to the control's. Its p-value is empirical, from the placebo windows.
• Fall. Monthly means for September 2025–August 2026 (Post12) against September 2023–August 2024 (Pre12). Pre12 ends before both Decreto-ley 9/2024 and the publication of LO 1/2025. Reported for all dwellings, for each kind, and as a ratio of ratios between kinds.
• Timing. Every month from January 2025 against the same month of Pre12.

§9 of METHOD.md lists what we added after the freeze:

• the historical list;
• the contrast for September 2024–February 2025;
• the run-up without the two cities with municipal suspensions;
• a correction to one confidence-interval formula.

After the first independent review we also added:

• placebo inference on spaced windows and placebo-calibrated intervals;
• the same calendar window in 2023, 2024 and 2026;
• older reference periods for the fall;
• the late-publication test of the date;
• the split of the unknown kind.

After the second review we added:

• the contrast within municipalities, now the primary flats-against-houses test;
• the excess by municipality, with València city and Torrevieja reported separately;
• the buildings and cadastral references of the València deadline cohort;
• the paired monthly comparison of flats and houses for the fall;
• the Generalitat's own year-on-year statement.

The spaced placebo set now leaves out the two June-2025 window ends whose baselines contain the run-up: 78 windows.

5. Results

5.1 What the date records

The dataset defines fecha_alta as «Fecha de alta en el registro» and signatura as «Número de inscripción en el registro». The data show that the date is set before the number. The filing date is the most plausible reading.

• Late publication with new numbers (a strong test).
  • The Internet Archive copy of 14 September 2026 ends on 11 September. Today's file has 25 dwellings that are absent from that copy but carry an earlier date. The 23 of them dated in 2026 all carry numbers above every number of their province already published on 14 September.
  • Numbers are therefore issued when a file is processed and published, and the date is set earlier.
  • This does not prove the date is the filing date. It could be another earlier date, such as the day the file was completed. But Ley 39/2015 art. 69.3 makes a declaration effective «desde el día de su presentación», which gives the office a reason to record exactly that.
  • Of the 108 dwellings now dated 1–11 September 2026, 96 were already in the 14 September copy; 12 appeared later.
  • In January 2025 the pattern repeats. 103 dwellings in the list of 24 January carry dates on or before 10 January but are missing from the list of 10 January.
• Processing order.
  • Each province numbers its dwellings in sequence: Alicante from 490000 since May 2022, Castellón and Valencia each on its own counter.
  • Of the 202 dwellings in the province of Valencia dated 2 April 2025, 62 were numbered after files dated more than a week later. The last of them was numbered after files dated 2 July 2025.
  • In Alicante the deadline cohort was processed faster: only 4 of its 62 dwellings waited more than a week. One was numbered after files dated 3 September 2025.
  • On its own, this argument would also fit numbers issued at filing and dates set at processing. The previous test argues against that reading.
• Non-working days.
  • 5.8% of the 38,933 registrations are dated on a weekend or holiday: 5.1 per Sunday against 40.1 per working day.
  • The share rose from 2.7% in 2023 to 5.7% in 2024 and 8.5% in 2025. That suggests more electronic filing, not office processing.
• Stability.
  • 72,863 of the 72,898 dwellings in both the January 2025 list and today's file (99.95%) have the same date.
  • Between 14 September and 4 October 2026, 30 dates changed. All moved earlier and all are in 2018–2019 cohorts, which reads as a clean-up of old records.

So a spike on 2 April is most plausibly a spike in filings, not a batch of processing. Processing took up to five months.

5.2 The rush

The region as a whole.

• Inference.
  • The 791 placebo windows overlap: their contrasts have a lag-1 autocorrelation of 0.93.
  • We therefore use every tenth window, leaving out the two whose baselines contain the run-up, which gives 78 windows.
  • None of them reaches the all-dwellings ratio (largest 1.47): p = 0.013.
• Seasonality. The same calendar window ran at 1.59 times its baseline in 2023 and 1.27 in
  1. Against their average, the 2025 excess is about 440 rather than 637, a ratio of 1.68.
• The single day. 2 April 2025, with 287 registrations, is the busiest day in the file since
  1. Without València and Alicante cities it has 114 and is the 9th busiest.

Where the excess is. 352 of the 637 excess registrations (55%) are in two municipalities.

• València city.
  • A deadline spike. The city ran at 4.43 times its expected count before 3 April and 0.32 times in the four weeks after.
  • The cohort of 2 April. 172 registrations, 140 of them units in buildings and 32 without a unit part, flat-like on our checks. They sit in 55 buildings by street and number; a stricter key that also separates blocks gives 76. 89 of the 172 are in buildings with five or more on that day, the largest with 19, 12, 10 and 10.
  • Missing cadastral references. 73 of the 172 have no cadastral reference, against 4.6% of the region's January–February 2025 registrations. Since Decreto-ley 9/2024 the reference has to be stated in the declaration.
  • Who filed. These are filings by multi-unit owners or operators, partly incomplete, as much as by individual owners.
  • The city's own rule. It had announced on 31 January 2025 that «No se podrá destinar a apartamento turístico un piso en una comunidad de vecinos». Its council approved the rule initially in April 2025 (§2.5). The city's spike therefore had two possible causes, and we cannot separate them.
• Torrevieja.
  • Its registrations rose from 10 March and stayed high after the deadline: 2.52 times the baseline in the four weeks after.
  • Its January–February 2025 baseline was low (76 and 132 a month, against 166 and 154 a year earlier).
  • That is a level shift, not a deadline rush. We found no municipal act to explain it, so we report it separately.
• The rest of the region.
  • 1.82 times the expected count, or 1.29 after the seasonal allowance: an excess of about 140.
  • Flats ran at 1.93 times and houses at 1.22, both within or near their seasonal range: in 2023 and 2024 the same window ran at 1.46 and 1.30 for flats, and 1.80 and 1.26 for houses.

Flats against houses.

• The within-municipality contrast is the primary test.
  • It compares flats with houses in the same municipality: a Poisson model with a municipality effect and a kind effect, with the expected counts of each municipality and kind as offsets.
  • In the same calendar window it was 0.82 in 2023, 1.10 in 2024 and 1.03 in 2026, so 1.73 is high for that time of year.
  • But 6 of the 78 spaced placebo windows are as large. The flats-against-houses difference is therefore not statistically clear.
• The pooled contrast of 2.02 is mostly composition. The two municipalities with the largest excess have almost no houses: València city had 0 houses in the window and Torrevieja 4.
• Counting filings by building and day lowers the pooled contrast to 1.81. The rush is not only a few multi-unit filings, but they matter.
• The unknown kind. Its 59 registrations in the window are flat-like: 50 are in València city and 52 have a cadastral unit other than 0001. Counting them as exposed or as control changes the pooled contrast to 2.04 or 1.50.

After the deadline.

• In the 28 days after 2 April, all dwellings ran at 0.92 of the baseline.
• That average mixes València city (0.32), Torrevieja (2.52) and the rest of the region (0.80). In the rest, flats ran at 0.83 and houses at 0.71.
• Outside Torrevieja, the weeks after the deadline were slightly below the pre-deadline pace. That fits filings brought forward, but the dip is small.

Without Torrevieja, the main results are:

• all dwellings ran at 2.19 times the expected count in the window (1.56 after the seasonal allowance) and 0.73 in the four weeks after;
• flats ran at 2.31 times;
• the within-municipality contrast is 1.79 (p = 0.101).

The size of the rush.

• About 440 to 640 extra registrations, depending on the seasonal allowance: two or three weeks of normal flow, in a registry of about 90,000.
• Half of it is in València city and Torrevieja.
• The run-up window lost no more numbers to cancellation than its baseline did: its gap share is 3.4%, against 6.3%.

5.3 The fall and its timing

The ratio-of-ratios intervals add the two kinds' variances as if they were independent. Common seasonality makes the two series move together, so those intervals are too wide. The paired estimate fixes this:

• for each of the twelve calendar months, it takes the log of flats-to-houses registrations in 2025–26 minus the same log in the reference month;
• the average of those twelve differences gives the estimate, and their standard error gives the interval.

With t-quantiles for 11 degrees of freedom, the interval against September 2022–August 2023 is 0.69–0.99.

By province, the ratios are 0.47 (Alicante), 0.65 (Castellón) and 0.39 (Valencia).

The reference period matters.

• 2023–24 was a boom in registrations.
• Against calendar 2022 the post-rule pace is 0.88 of the earlier one. Older cohorts have lost more registrations to cancellation (81.5% of 2022 registrations survive), and with that correction the ratio is 0.72.
• The houses' share of new registrations was unusually low in 2023–24, which flatters the flat-house contrast. By half-year it ran:
  • 18.1% and 17.9% in 2022;
  • 16.1% and 13.9% in 2023;
  • 14.4% and 14.6% in 2024;
  • 12.0% and 15.5% in 2025;
  • 20.4% in 2026H1.
• Survival does not differ by kind in the older reference either: 85.1% for buildings and 83.9% for chalets among September 2022–August 2023 registrations. So the weaker contrast against that period is not an attrition artefact. The scout's

measure, the monthly mean from May 2025 to August 2026 against 2024, gives 598 against 1,075 (a ratio of 0.56). The scout's 46% fall ran to September 2026, which is still being filled in.

Month by month, against the same month of Pre12:

The fall has three phases.

1. Before the run-up, flats and houses moved together. From September 2024 to February 2025, after Decreto-ley 9/2024 and before the rush, both ran at about 0.89 of a year earlier: 0.89 for flats and 0.90 for houses, a contrast of 0.99 (CI 0.79–1.25).

That decree had already required, since August 2024, a certificate that the community's statutes do not forbid the use. It did not, on this evidence, cut flats more than houses.

1. May–August 2025: the common fall. Both kinds fell to about two thirds of a year earlier: flats 0.66 (2,769 against 4,217), houses 0.65 (485 against 742). Something reduced registrations of every kind. The rule cannot be it, since it does not reach houses.
2. From October 2025: the gap. In September 2025 houses still fell more (0.46 against 0.53). From October, flats fell further, to 0.28–0.60 of the reference month, while houses stayed at 0.44–0.84. The whole-parcel share rose to about a fifth.

A modest flats-specific fall may exist. Paired by calendar month, flats' odds against houses in September 2025–August 2026 were 0.70 (0.62–0.80) of September 2023–August 2024 and 0.83 (0.71–0.97) of September 2022–August 2023. Against calendar 2022 the figure is 0.92 (0.82–1.03).

We do not attribute it to the rule. Its size depends on the reference period, because the houses' share drifts. Its timing fits no single enforcement step:

• the registrar's check of community agreements applied to national-registry applications from January 2025;
• that check was annulled by judgments of May–June 2026, published in June–July, yet the gap persists;
• the Valencian declaration began to ask for the community's express approval only on 3 July 2026.

5.4 Survivorship

The file keeps only live entries. A dwelling registered in 2024 and cancelled in 2025 is missing from the pre-period, and the pre-period has had longer to lose dwellings. Three measures bound the bias.

• The historical list. For Pre12, the list shows 14,177 registrations: those in the list of 10 January 2025, including those cancelled before then, plus those that reached the registry later. Today's file still holds 12,941 of them (91.3%).
  • Buildings (types A and B) survive at 91.1%, chalets (type C) at 92.3%.
  • Correcting Pre12 deepens the fall from 0.48 to 0.44.
  • It moves the contrast between kinds from 0.69 to 0.68.
• Number gaps. A number that is missing between two numbers present is either a cancelled dwelling, a file that was refused or withdrawn, or a number never used. Gaps are therefore an upper bound on cancellations.
  • Pre12 has 1,688 gaps. That is more than the 1,236 Pre12 registrations missing from today's file (425 of them already cancelled by 10 January 2025), as an upper bound should be.
  • The older Benidorm series (255 Pre12 registrations) cannot be linked to today's numbers. Put back on both sides, it gives a survival of 91.1% and a corrected ratio of 0.43.
  • Post12 has 106. If every one of them were a cancelled 2025–26 dwelling, the fall would still be 0.48.
• Removals between copies. Between 14 September and 4 October 2026, 248 dwellings left the file, about 5% a year. Of the dwellings registered in 2025–2026, 19 of 14,228 left.

The mass cancellations of 2025 reported in the press (886 in March, 10,601 in June) fell, by the press's account, on old records lacking a tax id or a cadastral reference. Their size is consistent with these bounds. Survivorship makes the measured fall smaller than the true one, not larger. It cannot create a one-day spike.

5.5 Other regions

The other registries also peak just before the deadline. Their date fields are undocumented, and they behave like office-processing dates. So they show that files were processed in bulk around the deadline, not when owners filed.

• Andalucía.
  • In the Registro de Turismo de Andalucía (OpenRTA, CC BY 4.0), registrations of tourist dwellings peaked on Monday 31 March 2025 at 787, the highest since 2016. Then came 688 and 599 on 1–2 April, and 13 to 71 a day from 5 to 10 April.
  • In the 14 days to 2 April there were 3,955 against 838 expected.
  • There were none on the weekend of 29–30 March, against 60 and 50 a week earlier. There were 220 and 409 on 3–4 April, after the deadline. activity_start_date equals registration_date for 71% of dwellings.
  • On the working days of 7–30 April 2025 the median was 31, against 76 in January–February.
  • Its export has misaligned rows. We recovered all 153,496 tourist dwellings by reading the named columns from the right end of each row (METHOD.md §2 and scripts/comparators/).
• País Vasco. 0.6% of its registrations fall on non-working days. The numbers are small, and the shape is the same.
• Mallorca. It has capped new places, and only three dwellings were registered there in all of 2025.
• Catalonia. The LPH does not apply there, so it would have been the outside control. Its open registry has no date. Whether the rush is absent where the LPH does not apply could not be tested. Inside the Valencian registry, the dwellings the rule does not reach serve as the control instead.

INE's experimental statistic counts listings on three platforms, not registrations. From May 2025 to May 2026 it fell by 18.9% in the Comunitat Valenciana (63,190 to 51,268), by 9.8% in Catalonia and by 10.7% in Spain. That is consistent with the registry's direction, but listings move for other reasons too: the national registry and the platform delistings ordered in 2025.

5.6 Our counts against the GVA's published totals

On 8 August 2025 the Generalitat said 8,579 dwellings had been registered between 8 August 2024 and 8 August 2025. It also gave 4,688 for January–July 2025, by province 3,391 (Alicante), 1,506 (Valencia) and 791 (Castellón), which add up to 5,688. Today's file holds more:

• 11,237 dwellings dated in the first period;
• 7,428 dated January–July 2025 (4,265, 1,814 and 1,349);
• 7,409 of those 7,428 numbered before 8 August 2025.

The Generalitat's statement also contradicts the file on direction. Europa Press reported it on 8 August 2025: «Desde la entrada en vigor del decreto, un total de 8.579 nuevas viviendas turísticas se han inscrito en el registro autonómico, lo que representa casi el doble que en el periodo anterior».

The file says the opposite:

• The year before. 13,451 dwellings dated 8 August 2023–7 August 2024 are still in today's file, and 14,528 were registered in that year according to the historical list.
• The year after. 11,237 dwellings are dated 8 August 2024–8 August 2025.
• The ratio is 0.84 on surviving dwellings and 0.77 against all registrations of the year before, not about 2.

Whatever the Generalitat counted, its "new dwellings" are not the file's dates of registration.

The processing lag does not explain the difference in levels. We do not know how the Generalitat counted. Possibilities include counting dwellings rather than declarations (a change of owner needs a new declaration since Decreto-ley 9/2024), or counting by inscription date. A count of registrations may therefore include re-registrations of existing dwellings. Re-registrations after a change of owner became compulsory only in August 2024, so they would inflate Post12 rather than Pre12.

6. Discussion

There was a registration rush before 3 April 2025. In the two weeks before the community-approval rule took effect, the Valencian registry received 2.4 times its expected filings, or about 1.7 times after allowing for the usual late-March rise.

But it was concentrated, and its causes are mixed.

• Where. 55% of the excess is in València city and Torrevieja.
• València city. It was preparing its own ban on tourist flats in residential buildings, announced on 31 January 2025. Its deadline-day cohort is dominated by multi-unit filings, many without a cadastral reference.
• Torrevieja. Its rise ran on past the deadline.
• Elsewhere. The rise was modest: about 1.3 times after the seasonal allowance.
• Flats against houses. Within municipalities, flats rose more than houses (1.73), as the LPH mechanism predicts. But the difference does not stand out from the placebo windows (p = 0.089).
• Another trigger. Some readers had been told that 3 April was a registration deadline.

So the data are consistent with a rush driven partly by the LPH rule. They do not let us attribute the rush to the rule alone. The single-day record and the pooled flats-against-houses contrast reflect València's municipal situation as much as the national rule.

The rush was also small next to the notice given. About 440 to 640 extra filings, two or three weeks of normal flow, after three months' notice. Most owners who could have filed early did not. As in French and Vreugdenhil's permit rush, filings that lock in the old regime need not mean new activity. In Spain, moreover, the protection may require that the owner was already letting.

Did the pace fall afterwards?

• Yes, against the two years before the rule. Registrations halved, from 13,154 to 6,253 a year.
• Most of the fall is shared. It is common to flats and houses and starts in May 2025, so the rule cannot explain it.
• It undoes a boom. Against 2022 the post-rule pace is 0.88 of the earlier one, or 0.72 corrected for cancellations.
• A modest flats-specific fall may exist (paired odds 0.83 against 2022–23). Its size depends on the reference, and its timing fits no single measure, so we do not attribute it to the rule.
• The Generalitat described the year after August 2024 as nearly double the year before. The file shows a fall.

What the data do not say: whether flats were stopped by community votes or by the expectation of them, or whether registrations that did not happen moved to seasonal lets or to unregistered supply.

7. Limitations

• Survivors only. The open file has live entries only. The historical list bounds the pre-period well, but cancellations after January 2025 are known only through number gaps and one archived copy.
• Registrations are not new dwellings. Some entries are new declarations for existing dwellings. The GVA's own totals for 2025 are lower than ours, and we cannot reconcile them.
• Kind classification.
  • It relies on the cadastral notation in the address, and it is stable over registration years (the share with unit number 0001 stays at 3–5% for in-building and 98–100% for whole-parcel dwellings).
  • 1.3% of dwellings have no unit part. 570 of those 1,211 were registered from 2025 on, 194 of them in the city of València.
  • Units in complexes count as exposed, as the LPH treats them. Some whole-parcel houses in private estates may still be subject to community rules, which would bias the contrast towards zero.
  • Survival by kind is measured with the GVA's building type (A and B against C), which matches our classification only approximately.
• Geography.
  • Flats and houses sit in different municipalities. The pooled contrast is therefore confounded by municipal events, and only the within-municipality contrast addresses that.
  • That contrast rests on 54 houses in the window across 35 municipalities.
• Inference.
  • Placebo windows overlap. Spaced windows (78) and placebo-calibrated intervals address this, but the smallest attainable p is then about 0.013.
  • The p-value without the two cities varies across the ten possible offsets of the spacing; we report the offset that starts at the first window.
  • The confidence intervals of the twelve-month ratios include seasonality.
  • The ratio-of-ratios intervals for flats against houses treat the two kinds as independent and are too wide. The paired monthly estimate is the one to read.
• Buildings. The building count (55 by street and number for the València cohort) depends on how addresses are matched. A stricter key that separates blocks gives 76.
• The plan and the reference period. The plan was frozen after the pilot was known. Its reference period for the fall, September 2023–August 2024, turned out to be a boom and a low point for houses' share. We report the older references beside it.
• Municipal measures. Their dates come from municipal notes and press, not from the provincial bulletins. The València dates of 31 January and 30 April 2025 are quoted from the city's own site. We found no municipal act of València on 2 or 3 April 2025, and none in Torrevieja to explain its rise.
• Case law. The Supreme Court's plenary rulings of October 2024 on «limitar o condicionar» in art. 17.12 were suggested to us. We could not read them in CENDOJ, so we do not cite them.
• Comparators. The Andalusian and Basque dates are undocumented and look like processing dates. Catalonia (no date), Madrid and Aragón (robots.txt) could not be used.

8. Data, code and licences

• Code: Apache-2.0. Our data and text: CC BY 4.0.
• data/ holds only aggregates: counts per day, province and kind; per month, municipality and kind; and every measure.
• The registries are not redistributed. Credits:
  • Generalitat Valenciana (Creative Commons Attribution);
  • Junta de Andalucía (CC BY 4.0);
  • Consell de Mallorca (Creative Commons Attribution);
  • Gobierno Vasco (portal legal notice, not read);
  • Instituto Nacional de Estadística ("Elaboración propia con datos extraídos del sitio web del INE: www.ine.es").
• Everything runs with the Python standard library: bash scripts/run.sh fetch, then bash scripts/run.sh.
• Code and data: https://github.com/easybytehub/easyxlab (studies/s13-tourist-registration-rush).

9. Automation and review

This study was carried out by AI agents (Claude, Anthropic), coordinated by an agent that wrote the analysis, under the direction of EasyByte's researchers. Specifically:

• The coordinating agent:
  • downloaded and filtered the Valencian registry;
  • designed the dwelling-kind classification;
  • wrote and froze the analysis plan;
  • wrote the code, tests and text.
• Three agents ran in parallel, each with a written brief and a budget of tool calls, using the same rate-limited fetcher:
  • one compiled the legal baseline from the BOE and the DOGV, with literal quotes;
  • one surveyed the other regions' registries and repaired the Andalusian export;
  • one ran the prior-work search.
• The coordinating agent checked their outputs against the sources it could re-read. It rebuilt the holiday list and the municipality dictionary with its own scripts, and both matched.
• No step was done by hand. scripts/check_headlines.py recomputes every headline number from data/ and checks that it appears in this paper and in the README.
• An independent agent then reviewed the study adversarially, without access to its drafts' reasoning. It recomputed about 30 numbers from data/. It found:
  • that we had dated the flat-specific gap a month early in one place;
  • that the causal reading of the later fall was not supported;
  • that the placebo p-values ignored the overlap between windows;
  • that the run-up excess ignored seasonality;
  • that the single-day record is mostly the city of València;
  • that the comparator dates look administrative.

We added the analyses that address those points (METHOD.md §9.9) and rewrote the claims.

• A second independent agent, launched by the coordinating researcher, reviewed that version. It recomputed the headline numbers with its own code and read the legal texts and the city of València's notes. It found:
  • that the flats-against-houses contrast was largely composition across municipalities (one blocker);
  • that Torrevieja's rise was not a deadline effect;
  • that València's own draft ban had been left out;
  • that a reference was wrong;
  • that real registry numbers appeared in code and small cells in a published table;
  • that the package could not be rerun from public sources.

This version applies those fixes (METHOD.md §9.10), including a new central claim.

• A partner of EasyByte has not yet read the study.

10. Competing interests

EasyxLab is the research lab of EasyByte Hub S. Coop. Mad., a software company. EasyByte has no client in tourist rentals, property management or registries, and no stake in the outcome. No one funded this study.

11. Ethics

• Registries of tourist dwellings can identify owners. We kept no names, addresses, cadastral references or registration numbers in anything we publish, and dropped them while reading.
• No published table below province level has a cell under 5: in monthly counts by municipality, smaller municipalities are merged into a provincial row (METHOD.md §8).
• The code's examples use synthetic registration numbers that do not exist in the registry.
• We respected robots.txt wherever it can work as a reservation of text and data mining, and never bypassed an access control. The one exception is the arXiv API, used under its terms.

How to cite

EasyxLab (2026). Did owners rush to register tourist flats before Spain's community-approval rule of 3 April 2025? Evidence from the Comunitat Valenciana's open registry. Study S13. EasyByte Hub S. Coop. Mad. https://github.com/easybytehub/easyxlab

References

• Ley Orgánica 1/2025, de 2 de enero, de medidas en materia de eficiencia del Servicio Público de Justicia. BOE-A-2025-76. https://www.boe.es/buscar/act.php?id=BOE-A-2025-76
• Ley 49/1960, de 21 de julio, sobre propiedad horizontal (consolidated). BOE-A-1960-10906. https://www.boe.es/buscar/act.php?id=BOE-A-1960-10906
• Real Decreto 1312/2024, de 23 de diciembre (Registro Único de Arrendamientos). BOE-A-2024-26931. https://www.boe.es/buscar/act.php?id=BOE-A-2024-26931
• Ley 15/2018, de 7 de junio, de turismo, ocio y hospitalidad de la Comunitat Valenciana. BOE-A-2018-8950. https://www.boe.es/buscar/act.php?id=BOE-A-2018-8950
• Decreto 10/2021, de 22 de enero, del Consell (alojamiento turístico), consolidated version. https://dogv.gva.es/dogv-portal/disposicion/355539/textoVigencia?idConsolidacion=4446&lang=es_es
• Decreto-ley 9/2024, de 2 de agosto, del Consell. DOGV of 7 August 2024.
• Ley 3/2026, de 29 de junio, de la Generalitat, art. 133. BOE-A-2026-15683.
• Ley 39/2015, de 1 de octubre, art. 69.3. BOE-A-2015-10565.
• Llei 5/2006, de 10 de maig, llibre cinquè del Codi civil de Catalunya. BOE-A-2006-11130.
• Ministerio de Vivienda y Agenda Urbana, press note of 2 June 2025. https://www.lamoncloa.gob.es/serviciosdeprensa/notasprensa/vivienda-agenda-urbana/Paginas/2025/020625-campana-registro-unico-alquileres.aspx
• Generalitat Valenciana. Datos de turismo sobre viviendas de uso turístico en la Comunitat Valenciana. https://dadesobertes.gva.es/es/dataset/tur-gestur-vt
• Generalitat Valenciana. Datos de turismo sobre viviendas de uso turístico en la Comunidad Valenciana 2025. https://dadesobertes.gva.es/es/dataset/dades-turisme-habitatges-comunitat-valenciana-2025
• Junta de Andalucía. OpenRTA. https://datos.juntadeandalucia.es/api/v0/openrta/all?format=csv
• INE. Viviendas turísticas en España (estadística experimental). https://www.ine.es/experimental/viv_turistica/experimental_viv_turistica.htm
• Barbosa, Pinto Machado and Terto (2026). Suspending Short-Term Rental Entry: The Portuguese Registration Ban and Its Reversal. SSRN. https://doi.org/10.2139/ssrn.7229455
• Guha, Meggiorin and Belo (2026). Before the Law takes Effect: Provider Anticipatory Behavior. SSRN. https://doi.org/10.2139/ssrn.7073098
• French and Vreugdenhil (2026). Permits as Real Options: Anticipation and Regulatory Leakage. NBER WP 35794. https://doi.org/10.3386/w35794 (abstract read at https://www.nber.org/papers/w35794)
• Shavell (2007). On Optimal Legal Change, Past Behavior, and Grandfathering. NBER WP 13563. https://doi.org/10.3386/w13563
• Kleven (2016). Bunching. Annual Review of Economics. https://doi.org/10.1146/annurev-economics-080315-015234
• Kleven and Waseem (2013). Using Notches to Uncover Optimization Frictions and Structural Elasticities. QJE. https://doi.org/10.1093/qje/qjt004
• Pons Esteva (2025). Análisis espacio-temporal del proceso de regularización de las estancias turísticas en viviendas: Mallorca (2016-2022). BAGE. https://doi.org/10.21138/bage.3529
• Garcia-López, Jofre-Monseny, Martínez-Mazza and Segú (2020). Do short-term rental platforms affect housing markets? Evidence from Airbnb in Barcelona. J. Urban Econ. https://doi.org/10.1016/j.jue.2020.103278
• Jin, Wagman and Zhong (2024). The Effects of Short-Term Rental Regulation: Insights from Chicago. NBER WP 32537. https://doi.org/10.3386/w32537
• arXiv 2211.16617 (2022). Identification of the Breach of Short-term Rental Regulations in Irish Rent Pressure Zones. https://arxiv.org/abs/2211.16617
• Terra Meridiana (21 April 2025). Housing chaos continues as tourist rental law comes into force. https://www.terrameridiana.com/36633-housing-chaos-continues-as-tourist-rental-law-comes-into-force.html
• Ajuntament de València (31 January 2025). Note on the draft rules for tourist dwellings. https://www.valencia.es/cas/actualidad/-/content/normativa-viviendas-tur%C3%ADsticas-val%C3%A8ncia
• Ajuntament de València (20 March 2026). Note on the definitive approval of the rules for tourist apartments. https://www.valencia.es/cas/actualidad/-/content/normativa-apartamentos-tur%C3%ADsticos
• elperiodic.com (9 June 2025). La Generalitat elimina 10.601 viviendas turísticas del registro. https://www.elperiodic.com/generalitat-elimina-10601-viviendas-turisticas-registro-turismo-contar-referencia-catastral_1021077
• Europa Press via Bolsamanía (8 August 2025). La Generalitat valenciana retira 12.000 viviendas turísticas del registro en un año. https://www.bolsamania.com/noticias_print/empresas/economia--la-generalitat-valenciana-retira-12000-viviendas-turisticas-del-registro-en-un-ano-y-tramita-7000-bajas-mas--20595974.html
• EFE via Alicante Plaza (29 June 2025). El registro único de alquileres de corta duración, de efectiva aplicación este 1 de julio. https://alicanteplaza.es/alicanteplaza/comunitat-valenciana/el-registro-unico-de-alquileres-de-corta-duracion-de-efectiva-aplicacion-este-1-de-julio
• elDiario.es Canarias Ahora (3 April 2025). https://www.eldiario.es/canariasahora/sociedad/ano-borrador-ley-canaria-alquiler-vacacional-norma-13-000-pisos-turisticos_1_12186625.html

Cite this study

EasyxLab (2026). Was there a rush to register tourist dwellings before Spain's community-approval rule of 3 April 2025? Evidence from the Comunitat Valenciana's open registry. Study S13. EasyByte Hub S. Coop. Mad. https://github.com/easybytehub/easyxlab/tree/main/studies/s13-tourist-registration-rush

@techreport{easyxlab_s13,
  title       = {Was there a rush to register tourist dwellings before Spain's community-approval rule of 3 April 2025? Evidence from the Comunitat Valenciana's open registry},
  author      = {{EasyxLab}},
  institution = {EasyByte Hub S. Coop. Mad.},
  number      = {S13},
  year        = {2026},
  url         = {https://github.com/easybytehub/easyxlab/tree/main/studies/s13-tourist-registration-rush}
}