Short-Term Rental Yields and Regulation in San Miguel de Allende

$1,096

Median listing, monthly gross

AirROI, Jul 2025 to Jun 2026. Before every cost.

29.9-44%

Annual occupancy, four-source range

No booking-verified dataset exists for this market.

MXN 45,000

Municipal lodging permit, per fiscal year

Ley de Ingresos SMA 2026, Art. 24 fracc. XIV.

16%

IVA on a furnished letting

Ley del IVA, Art. 20 fracc. II. Sits above the 4% state lodging tax.

The median short-term rental listing in San Miguel de Allende grossed USD 1,096 a month in the year to June 2026. The top decile grossed USD 5,242 or more. That is a ratio of nearly twelve to one, and it is the single most important fact about this market, because almost every yield figure quoted to prospective buyers describes the top of that distribution rather than its middle.

This page publishes the whole distribution, the four-way disagreement about occupancy, the full statutory cost stack, and the arithmetic of running a median listing rather than a fantasy one. Where a number is not publicly knowable, it says so.

The revenue distribution

AirROI publishes the only quartile breakdown available for this market. Annualised figures below are simply the monthly figures multiplied by twelve.

The median sits about 23 per cent below the mean. That gap is not noise. It is the arithmetic signature of a market where a professionalised minority captures most of the revenue, and it is why quoting the average annual revenue per listing, USD 17,033, as though it described a typical outcome materially overstates what a new entrant should expect.

Airbtics, working from a different window of February 2025 to January 2026, reports a median of MXN 309,000. That is far above AirROI’s implied median of roughly USD 13,150 and the two cannot be reconciled from published information. Airbtics’ own page is internally inconsistent on its dollar conversions, so the peso figure is the more defensible one to cite from that source. The same source grades San Miguel in the lowest three per cent nationally for short-term rental yield, with revenue up 5.9 per cent year on year but down 11.5 per cent over three years. That grade is striking and counter-narrative, which is precisely why it deserves scrutiny rather than amplification: Airbtics does not publish the methodology behind it, and yield rankings depend entirely on the property-price denominator chosen.

Every figure above is gross. None of it nets the 16 per cent IVA, management, tax, or the municipal permit.

Occupancy: a fourteen-point disagreement

Four scrapers, a fourteen-point spread on the number that determines whether a listing is viable. None of them publishes how it treats blocked calendars or dormant listings, which is almost certainly the driver: excluding unavailable nights from the denominator raises apparent occupancy. The honest position is a range of roughly 30 to 44 per cent, not a point estimate.

Two caveats on the table. The AirDNA figures could not be retrieved directly; that page returns HTTP 403, and the numbers come from search snippets rather than a page we read. GuestFavorites publishes revenue tiers in which the median exceeds the upper quartile, which is impossible, so its revenue data is excluded here entirely and its occupancy and listing count appear only as range context.

The state government’s count of 1,843 is not measuring the same thing as the scrapers. It is a count of properties the state can identify and attribute, which is the enforcement gap rather than the market.

AirROI’s internal consistency is worth noting in its favour: it reports RevPAR of USD 64, and 29.9 per cent occupancy against a USD 210 average daily rate gives USD 62.80. The rates themselves also cluster more tightly than the raw table suggests. Converted at the 23 July 2026 reference rate of 17.46 pesos per dollar, GuestFavorites’ MXN 3,680 is about USD 211, close to AirROI’s USD 210, with AirDNA’s USD 178 nearby. Airbtics at roughly USD 110 is the outlier, and it is also the source reporting the highest occupancy, which is the opposite corner of the same trade-off.

The benchmark that makes 30 to 44 per cent look normal

SECTUR’s monitored hotel occupancy for San Miguel de Allende was 45.7 per cent in week 9 of 2024, one of the strongest weeks of the year, against a statewide average of 61.8 per cent across twelve monitored destinations. Statewide occupancy in January 2026 was reported at 33 per cent. San Miguel underperforms its own state on hotel occupancy in the data available. Short-term rental occupancy in the low thirties to low forties is therefore structurally plausible rather than anomalous.

The figures that circulate in local marketing, hotel occupancy rising from 90 to above 98 per cent, describe a single Día de Muertos week. Presenting a festival week as an annual rate is the most common and most misleading move in this market.

Seasonality

Seasonality here is an occupancy phenomenon, not a pricing one. Occupancy swings by a factor of 1.8 across the year while the rate moves by only 1.2. Owners who assume they can price their way through the summer trough are working against the shape of the demand curve.

AirROI groups February, March and December as peak season, averaging 38.4 per cent occupancy, and May, June and September as low season, averaging 26.2 per cent.

There is a live conflict on which month leads. AirROI puts February first; GuestFavorites identifies March and April as peak and September and October as low. Both cannot be fully right. Semana Santa fell between 22 March and 8 April in 2026, and its movement between March and April year to year probably explains part of the disagreement. The widely repeated local claim that March is unambiguously the peak month is not what the one dataset publishing monthly figures shows.

December is named as a peak month but no source publishes a December-specific average daily rate. Only February’s USD 222 and June’s USD 183 are available at month level. Any December rate quoted to you is either an inference or a misread of the annual market average.

The incumbent base

A 46-day lead time marks San Miguel as a planned-trip destination rather than a spontaneous one, which matters for how quickly a new listing can fill a calendar. Half the listings hold Superhost status. A new entrant is not competing against an amateur field, and should not assume it will reach market-average occupancy in year one.

The cost stack

Three separate layers of public charge apply, and they are frequently conflated with one another.

The IVA line is the one most often left out, including by every published cost stack for this market we could find. Letting a dwelling is exempt from IVA. But Article 20 fracción II of the Ley del IVA withdraws that exemption for property supplied furnished, or destined for or used as a hotel or lodging house, so a furnished short-term let is taxable at the standard 16 per cent however short or long the lease runs. That is a statutory distinction rather than an enforcement position, and at 16 per cent of gross it is the largest tax charge in the stack. For a host holding an RFC the platform withholds half of it at source under the digital platforms regime; the other half remains the host’s own to declare.

The rate and the liability come from the statute. The withholding mechanics above them, the 4 per cent ISR and the 8 per cent IVA taken at source, are sourced from Mexican tax advisory publishers rather than from the statute text directly, and the penalty rates that apply without an RFC should be verified against the law before anyone relies on them.

Management sits on top. The most concrete published figure is 20 per cent of the rental income stream, from a named San Miguel operator, Berkshire Hathaway HomeServices Colonial Homes, which is also a brokerage selling into this market. A commission-adjacent publisher gives a prevailing range of 20 to 30 per cent for short-term rentals against 8 to 12 per cent for long-term, and a narrower 18 to 25 per cent elsewhere on the same site. The 20 per cent floor is corroborated; the 30 per cent ceiling is not.

Modelling a median listing

Take the median listing, USD 13,152 of annual gross, an RFC-registered host charging IVA at the statutory 16 per cent on gross with no input credit taken, management at 20 per cent, and a property with a fiscal value of MXN 5,000,000 paying predial in January to capture the 15 per cent early-payment discount. Fiscal value is not market value, and the ratio between the two is not published for San Miguel, so read the predial line as an order of magnitude rather than a quote. That 20 per cent is the only management rate in this market published by a named operator; the wider 20 to 30 per cent range is uncorroborated, so the model uses the figure with a publisher’s name on it rather than the midpoint of a range nobody stands behind.

That remaining USD 3,982 has not yet paid for electricity, gas, water, internet, cleaning consumables, linen, civil liability insurance, repairs, furniture replacement, accounting, or any period the owner uses the property personally. It also assumes full compliance with a permit most operators do not hold.

Two qualifications sit on the IVA line, and they cut in opposite directions. The model charges the full 16 per cent and claims nothing back, but the IVA a host pays on the management fee is creditable input tax; at 16 per cent of the USD 2,630 fee that is about USD 421 of credit this model does not net off. Against that, none of the four scrapers discloses whether its revenue figures are quoted before or after the tax, so the USD 13,152 of gross may already be stated net of it. Neither point is resolvable from anything published, and the model takes the more conservative reading of each.

The model is worth cross-checking against the only sourced attempt at a full cost stack. A commission-adjacent publisher puts combined operating costs at 55 to 65 per cent of gross. The lines above come to 69.7 per cent while excluding utilities, cleaning and insurance entirely, so the statutory and management stack alone exceeds the top of that published band before a single operating expense is added. That publisher’s stack appears to deduct neither the municipal permit nor the IVA, which accounts for most of the distance. Substituting its 60 per cent midpoint for the itemised lines would leave about USD 5,300 rather than USD 3,982, and the gap between those two figures is a fair measure of what the published one leaves out.

The same source puts break-even occupancy at roughly 25 to 30 per cent, a band that brackets AirROI’s measured 29.9 per cent. Read that as a hypothesis rather than a finding. It rests on an undisclosed methodology, and it comes from the publisher whose cost stack omits both the municipal permit and the IVA, which would put the true break-even above the range it gives. It does nonetheless align with the independent assessment placing this market in the lowest three per cent nationally for yield.

The permit is regressive

Because the MXN 45,000 charge is flat rather than proportional, it lands very unevenly.

A compliant bottom-quartile listing hands nearly half its gross revenue to the municipality before it pays anyone to clean it. Run the whole stack across that tier and the position is worse than regressive. USD 5,352 of gross carries USD 856 of IVA, USD 1,070 of management, USD 428 of state and federal withholding, the USD 2,577 permit and USD 807 of predial, which is USD 5,738 of cost against USD 5,352 of revenue: a loss of USD 386 before a peso of maintenance. This is the mechanism by which the fee, whatever its intent, pushes the small and occasional operator either out of the market or into non-compliance. In year one, adding the MXN 11,248.64 change-of-use charge takes the municipal burden on a median listing to about USD 3,221, or 24.5 per cent of gross.

Why this page publishes no net yield percentage

A yield needs a denominator, and there is no defensible one. Every San Miguel transaction figure available traces to a single brokerage’s own MLS view; there is no public registry of sales. Published net yield figures for this market all come from commission-adjacent sources with undisclosed methodology, and at least one of them appears to deduct neither the municipal permit nor the IVA.

What can be said is arithmetic. For the USD 3,982 above to represent a four per cent net return, the property would have to have cost about USD 99,500. Readers can substitute their own acquisition price and draw their own conclusion.

The regulatory position

Guanajuato has had a dedicated short-term rental statute since December 2020, the Ley de Hospedaje a través de Plataformas Digitales. Its substantive obligations are lighter than commonly assumed.

Hosts must register in the Registro de Prestación de Servicios de Hospedaje, held within the state tourism registry. Registration is free and done through online forms. Hosts must carry civil liability insurance covering the lodging risk whenever a reservation is active, and the statute permits that cover to be provided through the platform without naming any platform or setting a minimum amount. Hosts must hold an internal regulation, and where the unit sits within a condominium regime they must also exhibit the condominium’s governing instrument. That provision requires disclosure of condominium rules. It does not authorise short-term rental use, and it does not prohibit it. Whether a given building permits it is governed by its own escritura and reglamento, which are private documents.

Separately, under the state tax law, lodging providers must register with the state contributor registry within ten business days of starting operations, and file definitive monthly returns by the 22nd of the following month, including nil returns, until they formally deregister.

The statute carries no fine schedule of its own. Article 34 cross-references the state Tourism Law and its regulations for sanctions.

The clause that matters most

Platforms have been the statutory withholding agents for the 4 per cent lodging tax since a reform published on 30 December 2022. But Article 57 relieves the host of calculating and remitting only on the base collected through the platform, and only when the platform actually makes the payment. The relief is conditional, not automatic.

A June 2026 tax guide states that Guanajuato is not among the states where Airbnb handles lodging tax retention and direct payment. In July 2026 the state’s subsecretary of finance publicly asserted that platforms have not complied for three and a half years, that Airbnb refuses to let the state review its calculations, has no registered fiscal domicile in Guanajuato, and refuses to share host identity data. Airbnb has published no response on the record, and the reporting comes from a single outlet.

The practical consequence for a host is unambiguous even though the dispute is not. If the platform is not remitting, the conditional relief does not apply, and the 4 per cent is the host’s own liability. Income from direct bookings never had the relief in the first place.

Enforcement as it actually stands

The 2022 collection figure is the single best evidence that municipal enforcement has historically been weak, and it describes a calendar year that ended more than three and a half years ago. Nothing more recent is public.

The SATEG campaign is the live development, and its scope is narrower than the headlines suggest: it targets the 2021 to 2022 period, before liability shifted to platforms in December 2022. It is reported by one outlet and has not been independently confirmed.

Direction of travel

The escalation narrative is weaker than it sounds. The municipal fee went from MXN 10,000 in 2022 to MXN 45,000 in 2023, an increase of more than 300 per cent, and has not moved since. It was not increased for 2026. Guanajuato’s 4 per cent lodging tax survived the November 2025 reform of the state finance law unchanged, and sits mid-table nationally: the lowest states charge 2 per cent, the highest 5 to 6 per cent.

Pressure is coming from two directions instead. The first is collection rather than rates, visible in the SATEG campaign and the public dispute with Airbnb. The state originally projected MXN 30 to 40 million a year from taxing digital platforms when the law passed in 2020. It says that has substantially not materialised.

The second is the hotel lobby, which is an interested party and should be read as one. The president of the San Miguel hotel and motel association has said the state law “nació muerta porque no hay reglamento” and is therefore useless. He also estimates that Airbnb supply is comparable to the city’s roughly 3,500 hotel rooms, which is a judgement rather than a count from a direct commercial competitor. The underlying legal claim is independently plausible, however: the 2020 law required an implementing regulation within 120 days, and no evidence that one was ever published could be found.

An investor underwriting on rental income should model the fee as a recurring annual cost, assume the 4 per cent is their own liability until a platform statement says otherwise, and treat collection intensity, not rate changes, as the variable most likely to move.

What nobody publishes

These are gaps in the public record, not omissions from this page.

• The compliance rate. No 2026 municipal count of how many properties actually hold and pay the MXN 45,000 permit. Arguably the most important regulatory number for an investor, and it does not exist publicly.
• The fine schedule. The state statute cross-references the Tourism Law for penalties. The peso amounts, and the specific penalty for operating without the municipal permit, could not be sourced.
• Whether the implementing regulation exists. The 2020 law required one within 120 days. The hotel association says it does not exist. No official source confirms either way. This is an absence of evidence, not evidence of absence.
• Any booking-verified data. Every revenue, occupancy and rate figure here comes from listing scrapers or commission-adjacent publishers. There is no AMPI, MLS, or notario-association dataset for short-term rentals in this market, and no public transaction registry for the property prices that would form a yield denominator.
• Occupancy by revenue tier. The distribution is published for revenue only. Any claim that the top decile achieves a specific occupancy rate is not sourced to a published dataset.
• Breakdown by bedroom count, property type, or neighbourhood. Airbtics gates its bedroom-count figures behind a paid dashboard. No neighbourhood-level short-term rental data is published for San Miguel.
• Whole-home versus private-room split. Unknown, which materially affects how the revenue distribution should be read.
• Verified net yield after the full cost stack. Nothing published deducts IVA, management, the 4 per cent, federal withholding, the municipal permit and predial together with a disclosed method.
• Whether the IVA is enforced against individual hosts. That a furnished letting bears 16 per cent is statutory and not in doubt. Whether SAT actually pursues it against individual owners, most of them foreign, is undocumented: no collection figure, audit count or compliance rate for it has been published for San Miguel or for Guanajuato. Statutory liability and collected tax are different quantities, and only the first of them is knowable here.
• Whether scraped revenue is quoted before or after IVA. None of the four sources states it. If their figures are already net of the tax, the model above deducts it a second time and understates the result by roughly USD 2,100 at the median; if the figures are gross of it, the model is right. Nothing published settles which, and the whole net figure turns on it.
• Condominium-level restrictions. Governed by private escrituras and reglamentos. No survey of San Miguel condominium regimes exists.
• A 2026 hotel occupancy figure for San Miguel. The most recent official monitored figure is week 9 of 2024. OTEG’s monthly report downloads return a JavaScript shell rather than the document.
• Any Airbnb response to Guanajuato’s July 2026 allegations, and any independent confirmation that the platform does not withhold in this state.

All conversions on this page use 17.46 MXN per USD, the ECB reference rate for 23 July 2026. Annual average exchange rates are a different measure and are not used here. Past performance does not indicate future returns.