The Downside Case for San Miguel de Allende Real Estate

29.6 months

Average inventory, Jan-May 2026

Against 17.6 months a year earlier. The same source calls 15 to 18 months normal for this market.

12.6%

Fall in average sale price, 2025

$650,462 to $568,829. A change in what sold as much as a change in what things are worth.

12

Sales above $2 million, 2025

Down from 19 in 2024. The entire top of the market is a dozen transactions a year.

49.3

Homicides per 100,000, Sept 2024-Aug 2025

88 homicides in the municipality. Mexico's 50th most violent, on SESNSP and INEGI data.

The pages that rank for “will San Miguel de Allende real estate go down” are published by firms that earn a commission when a transaction closes, and the most sceptical result on the first page for the bubble query is a forum thread. That is not a conspiracy. It is a structural fact about who publishes.

We earn nothing on whether you buy, sell or do neither. That does not make this page correct. It makes it possible.

What the bear case is not

It is not a crash forecast, and the mechanism matters more than the sentiment.

Housing crashes are leverage events. Prices fall, borrowers go underwater, lenders force sales, forced sales set new comparables, and the loop tightens. San Miguel de Allende lacks the ingredients. Its principal data source describes the market as almost exclusively cash, without quantifying the share, and mortgage rates quoted to foreign borrowers in Mexico run around 10 to 14.5 percent, which prices most people out of financing entirely.

Carrying costs are correspondingly trivial. Predial runs on marginal rates of 0.234 to 0.354 percent of fiscal value, which is not market value and whose ratio to market value is published nowhere. An owner with no mortgage and a four-figure peso tax bill can wait years rather than accept a low offer.

So the downside here does not look like 2008. It looks like time.

The measured evidence as it stands

Every figure below comes from one brokerage reporting on its own view of a local MLS. No public registry publishes San Miguel transaction data, and AMPI San Miguel de Allende publishes none either.

The last row needs a warning label. The June year-to-date figure cannot be reconciled with the same firm’s May post, which reported 197 closings through May 2026 against 190 a year earlier, and a year-to-date count cannot fall from 197 at five months to 186 at six. Until that is reconciled, the 2026 transaction count for this city is not publicly known, and any page presenting one is presenting a choice rather than a fact. The 2025 count is a related trap: the source published 404 closings for 2024 and a 24.68 percent increase for 2025, but never an absolute number. Roughly 504 is implied. It has never been reported.

Risk one: the supply build

The same brokerage that publishes the inventory figure describes 12 or more months as ordinary for San Miguel and 15 to 18 as normal, on the argument that a cash market never absorbs like a mortgage-driven one. On its own benchmark, 29.6 months is roughly double normal. In March 2026 it reported inventory in several price bands at levels not seen since the period following the 2008 financial crisis, without publishing the bands or the counts.

The June fall from 860 listings to 745 is not absorption. The source attributes it to withdrawals, expirations and sellers choosing other strategies, and separately describes the decline as 81 properties when the arithmetic gives 115. Listings leaving because sellers gave up is a bearish datapoint in a bullish shape.

What it does not mean is distress. Under-contract dollar volume in March 2026 was 35 percent higher than a year earlier even as fewer properties went under contract, and transaction count in the May series was roughly flat while listings rose 45 percent. This is a supply-side event, not a demand collapse.

Risk two: you may not be able to get out

There is no days-on-market series for San Miguel de Allende from any transaction dataset. What exists is the shape of supply. One brokerage’s colonia directory, retrieved on 23 July 2026, shows 220 active listings in Centro against 13 in Ojo de Agua, 13 in Atascadero and 11 in Guadiana. Those are not markets in any statistical sense. They are handfuls of properties where a single sale is the entire comparable set. At the top it is starker: twelve sales above $2 million in the whole city in 2025 is roughly one a month. A seller at that level is not in a market. They are waiting for a particular person.

Then there is the round trip, which is where “can you lose money” is usually answered in practice.

On the published ranges, a round trip costs somewhere near 12 to 14 percent of value before any tax on gain. That is arithmetic on ranges rather than a measured figure, and the real number depends on what a buyer negotiates and how a notario assesses. It is still the honest starting point: a property must appreciate by roughly that much before the owner is level. The depreciation rule is the part most often omitted, and it works against long holds specifically.

Risk three: currency, which cuts in an unobvious direction

San Miguel prices are quoted and paid predominantly in dollars, as peer-reviewed work on the city also finds. So a stronger peso does not lower headline asking prices. What it lowers is purchasing power over everything denominated in pesos: renovation, labour, predial, notary fees, transfer taxes, maintenance. Spot USD/MXN strengthened through 2025 to close the year near 18.00 and stood at 17.46 on the ECB reference rate at 23 July 2026. A US$3,000 monthly transfer that converted to roughly 62,600 pesos at the start-2025 level implied by published figures now yields about 52,400.

Two distinctions get mangled constantly. Spot rates and annual averages are different measures: the IRS yearly average was 18.330 pesos per dollar for 2024 and 19.212 for 2025, describing the level across each year rather than the start-to-end move. And peso forecasts are not peso facts. The carry differential supporting the strong peso was Banxico at 7.0 percent against the Federal Reserve at 3.75 percent in March 2026, a compressing spread, and consensus forecasts for end-2026 cluster around 18.75 to 19.00.

The subtler exposure is on exit. Mexican capital gains are computed in pesos, with acquisition cost indexed for inflation. A property flat or down in dollars can still produce a taxable peso gain. How large depends on the peso path and the index over the holding period, which no published figure can settle in advance. It is a question for a Mexican tax adviser before signing, not after.

No source measures how much of the 2026 slowdown the currency explains. The purchasing-power arithmetic is solid; the causal link to transaction volumes is inference.

Risk four: the buyer is upstream, in another country

The most interesting explanation for the 2026 inventory build comes from the brokerage whose interest it least serves. Its February 2026 update attributes softness to the US housing market, citing roughly 147 sellers per 100 buyers nationally, the worst ratio since 2013; its May update cites US inventory up 27 percent over 21 months. The stated mechanism is that many San Miguel buyers must sell a US property first.

Treat those figures carefully. They are US national statistics used as context for a Mexican municipality, the firm names no data provider, and our fact-check could not verify either against a primary source. The mechanism is nonetheless plausible, because if it is right then San Miguel’s 2026 is being set in Phoenix and Denver rather than in Guanajuato.

The size of that exposure is unmeasurable. No source publishes a buyer nationality mix for San Miguel de Allende. The only official nationality data is the 2020 census, which recorded 1,100 arrivals from the United States over five years and an implausible 15 from Canada, and which counts residency rather than purchases.

Risk five: one industry, one demand source

San Miguel’s economy is tourism and the people tourism attracts. The mayor reported roughly 2 million visitors and about 14 billion pesos of economic impact for 2025, contradicted by a second outlet reporting 2.19 million visitors and 8.564 billion pesos. Neither is independently audited.

The municipal budget shows how levered local finances are to property changing hands. For fiscal 2026 it budgets MXN 324,352,284.67 from acquisition tax against MXN 210,859,321.49 from annual property tax, out of total budgeted income of MXN 1,300,127,987.83. Roughly 41 percent of budgeted income depends on property, and the larger half of that on transactions occurring rather than on properties existing.

The lodging picture is less buoyant than the visitor headline suggests. The most recent official monitored hotel occupancy figure for San Miguel is 45.7 percent in week 9 of 2024, against 61.8 percent across the state’s monitored destinations that week. Short-term rental occupancy estimates for the year to mid-2026 range from 29.9 to 44 percent depending on the scraper, and one grades the market in the lowest 3 percent nationally for yield, with revenue down 11.5 percent over three years. That grading discloses no methodology. Together the figures suggest rental supply outgrowing demand even as visitor volume rises.

Risk six: short-term rental regulation is a live enforcement story

The rates have been stable. The enforcement has not.

The municipal permit is a flat charge, which falls hardest on modest listings: the same MXN 45,000 whether a property grosses the market median or ten times it. It has not risen since 2023 and was not increased for 2026, a genuine counterpoint to any narrative of escalating municipal pressure.

The change is at state level. In July 2026 Guanajuato’s tax authority sent roughly 5,000 invitation letters to platform hosts, described as a friendly first phase with audits possible in a second, and stated publicly that platforms have not complied with the 4 percent lodging tax for three and a half years. A June 2026 tax guide indicates Airbnb does not withhold it in Guanajuato, which would leave the 4 percent with the host despite the platform being the statutory withholding agent since December 2022. Airbnb has published no response we could locate. Historic compliance was weak: the municipality collected MXN 175,442 from the lodging charge in all of 2022, from around 243 registered properties. Today’s compliance rate is published by nobody.

A property marketed on its rental income is often priced on a yield that quietly assumes the permit is not paid, the lodging tax is not remitted and the furnished-rental IVA does not apply. Enforcement does not need to change the law to change that price.

Risk seven: security, stated accurately

The reassuring version of this section is easy to write and it is wrong.

Guanajuato is at Level 3, reconsider travel, on the US advisory issued 29 May 2026, one of seven states at that level. Canada advises against non-essential travel south of and including highways 43D and 45D, and the UK FCDO against all but essential travel south-west of road 45D, naming 20 municipalities. San Miguel appears on none of those lists, and the current US advisory does not mention it at all. The sentence still quoted across brokerage sites, reassuring readers that no restrictions apply to San Miguel, is from a superseded version.

Guanajuato recorded 789 intentional homicides in the first half of 2026 on SESNSP figures, the highest of any state and roughly 8.8 percent of the national total, though down sharply year on year. An independent press tally counts 1,141 victims for the same period, a 45 percent gap arising from different methodologies. Both belong on the page.

San Miguel de Allende is far quieter than the industrial south: 3 homicides in June 2026 against 45 in León and 32 in Salamanca. But over the twelve months from September 2024 it recorded 88 homicides, a rate of 49.3 per 100,000, making it Mexico’s 50th most violent municipality. That window predates the 2026 statewide decline, so it describes a worse period than the present. It also refutes the “largely insulated” claim outright.

February 2026 is the case study. Mexican press reported an arson incident in San Miguel on 22 February with businesses and universities suspending activities, and a body showing signs of violence found in the municipality on 25 February, within a statewide reprisal wave following El Mencho’s death that reached roughly 23 Guanajuato municipalities. The brokerage supplying every price figure on this page reported that month as free of unusual incidents. For an owner this is tail risk transmitted through demand rather than title: advisories and headlines move sentiment in the countries the buyers come from.

Risk eight: water

CONAGUA’s 2024 determination for the Cuenca Alta del Río Laja gives recharge of 139.7 million cubic metres a year against registered extraction of 201.68 million, a deficit of 61,984,190 cubic metres annually, and states that no volume is available for new concessions. A well-drilling ban has covered the municipality since decrees of 1958 and 1964. This is the rare risk that is also a support, which is why it appears in both halves of the page.

The other side of the ledger

A balanced page states the supports with the same specificity as the risks.

Two deserve elaboration. The Centro constraint is federally enforced: any construction, restoration or conservation work inside the monuments zone requires INAH authorisation under the 1982 decree, separately from municipal permits. That cannot be arbitraged away, and it is the strongest structural argument for the walkable core specifically. It says nothing about the colonias where inventory has actually built.

The expat infrastructure argument is usually made in a lifestyle register and is sounder in a market one: a deep pool of long-established foreign residents and institutions is what lets a market re-liquify after a shock. It is weakest where buyers are most exposed. The municipality has 26 public-sector medical units, one IMSS and one ISSSTE among them, and complex care means a transfer to Querétaro.

What would have to be true for the bear case to bite

Each of these is observable, and the sources are named so they can be checked without us. This is the part a buyer can use.

The bull version is equally fair and shorter. Inventory falling back toward 15 to 18 months without price cuts, the peso weakening toward 19, the US housing market unfreezing, and the top tier recovering toward 19 sales a year would together describe a market that simply had a slow eighteen months. Nothing in the data rules that out.

The summary

A collapse is unlikely because the machinery that produces collapses is not installed here. A long, quiet repricing is plausible, is partly visible already, and would be almost invisible in asking prices while very visible in how long anything takes to sell.

The realistic way to lose money here is not a crash. It is paying an asking price set in a 17.6-month market, in a colonia with eleven listings and no comparables, in dollars, then needing to exit in a year when 29.6 months of supply competes for the same handful of buyers, after roughly 12 to 14 percent of round-trip cost and a peso-denominated tax bill on a dollar-flat outcome.

One aggregator wrote the fairest sentence any commercially interested party has published about this market: it is not a distressed market, but it is no longer a market where every property sells quickly at any price.

This is published research, not legal or tax advice. Your notario is the legal authority on your transaction, and Mexican capital gains treatment should be confirmed with a Mexican tax adviser before you sign anything. We have no financial interest in whether it closes.