Bali Property Investment in 2026: What the Published Numbers Say

Three numbers decide whether a Bali villa makes money: how many nights it is let, what each night earns, and how much of the purchase price is gone when the lease runs out. The published data gives a wide range for the first two and nothing at all for the third. Occupancy for the island in 2026 is 36%, 57% or 63% depending on which dataset you open. Gross yields worked from those datasets and the median sale price land between 4% and 8%, before a single cost. The 12% to 18% that the ranking pages quote is gross, for the best segment, at an occupancy no independent source shows for Bali.

What follows is the case for and against, built only from third-party figures with dates, and a list of what nobody publishes. The ownership routes, the areas and the taxes have their own articles and are linked where they come up.

Demand: the record year and the flat one

Bali’s case rests on tourism, and 2025 made it well. The statistics office counted 6,948,754 direct foreign arrivals, up 9.72% on 2024, with Australia alone supplying 23% of them. Star-hotel occupancy averaged 60.8% for the year. The luxury end did better: Horwath HTL and C9 Hotelworks put hotel occupancy at 73.2% in 2025, with a July peak of 85.9%.

Then 2026 stopped growing.

Source: BPS Bali monthly releases, January to August 2026, via the office’s own site and Antara. “Down” where the office reported a fall without the percentage; n/a where its release gave no comparison.

Two things in that table matter more than the headline. When the July figures came out, the statistics office said that in 2026 only February and July had risen year on year. August then added a third month, by half a percent. Five of eight fell. The provincial tourism office, for its part, set its 2026 target at 6.625 million in March, which is below what 2025 delivered. A target set below last year’s result forecasts a plateau. One more number from the same office: the IDR 150,000 tourist levy was paid by 34.8% of arrivals in 2025, IDR 369 billion against a budget of 500 billion, so the levy is no check on the count either.

Supply, meanwhile, kept coming. Horwath and C9 count 5,641 hotel rooms in the pipeline across 45 projects, more than half of them in the top two rate bands, and over 70 hospitality-managed residence projects on sale, with roughly 40% of them in Canggu and Berawa. In its second-quarter 2026 report Colliers described hotel occupancy as “normalising as new supply enters”, which in a year of flat arrivals and new rooms means lower. The visitor count that must absorb all of it has stopped growing.

Occupancy: three datasets, three answers

Every yield claim stands on an occupancy figure, and for Bali the published ones disagree with each other by a factor of nearly two.

Why the spread? Denominators. One service divides booked nights by the nights an owner made available, another by every night in the year, and a third drops listings that went quiet. AirDNA’s own page shows active listings down 45.7% in a year while occupancy rose 34.6%, which is what a change in sample looks like, not a change in Bali. None of the three publishes a full method; AirROI gives a one-line definition, the other two none.

Read the table for what the figures agree on. Nothing covering the whole island reaches 70%. The hotels that do, in the Horwath set, are managed luxury properties with sales teams. A listing page that assumes 70% to 85% for a two-bedroom villa is assuming the villa will outperform every published average, including the five-star hotels, which ran at 72% in June 2026. By area the lowest dataset is no kinder: AirROI has Canggu at 36%, Ubud at 37%, Seminyak at 38% and Pecatu, the best of its thirty sub-markets, at 40%.

Yield, worked from figures someone else published

Start with the number no agency quotes. Global Property Guide, which divides median asking rents by median asking prices from a listing portal, puts Bali’s gross yield on long lets at 4.24% as of August 2026: 3.3% for a one-bedroom, 5.5% for two, 5.2% for three, 2.9% for four or more. Its method is crude and it says so. It is also the only published Bali yield with a stated method and a fixed update cycle.

Short lets earn more. Take AirDNA’s USD 24,800 of annual revenue per listing and the USD 298,000 median sale price that REID reported for the first quarter of 2025 (as relayed by the Real Estate 3.0 newsletter in June 2025; REID’s own reports are not public), and the gross yield is 8.3%. Do the same with Airbtics’ median of USD 21,000 and it is 7.0%. Both are island-wide medians, so half of all listings earn less, and both are revenue before a single cost. Both services also show that revenue falling: Airbtics’ median down 15% in a year, AirDNA’s nightly rate down 19%, and REID’s own first-quarter note had rental revenue down 16%.

Now the costs that are fixed by law, on that USD 24,800, taking it as what guests paid.

• Hotel-services tax, 10% of the guest’s bill, collected by the operator and paid to the regency: USD 2,480, leaving USD 22,320 of receipts.
• Income tax for a non-resident owner, 20% of those receipts, withheld: USD 4,464. A resident on long lets pays 10%; on short stays the tax office treats the income as business income.
• Land and building tax in Badung, 0.2% of assessed value for a property assessed above IDR 1 billion, less a small allowance: around USD 590 on a villa assessed near the sale price.

That leaves USD 17,300, or 5.8% of the price, before anyone has cleaned a pool. The operator’s fee, the staff, the electricity, the pool chemicals, the furniture that a holiday let replaces every few years: for none of these does a survey exist. Agencies quote management as a share of receipts, and the quotes differ from one another by more than double, which tells you what agencies charge. What a villa costs to run is a different number, and nobody has published it. Two of its parts are on public tariffs: PLN’s business rate in October 2026 is IDR 1,445 per kWh, and Badung’s minimum wage for 2026 is IDR 3,791,002 a month per employee. At a 20% operator fee on receipts the figure drops to USD 12,800, about 4.3%. Staff, power, water and furniture come out of that.

The rates and who owes each of them are set out in Bali property taxes for foreigners. The point here is narrower. The gap between 8.3% gross and whatever is left is where every “12% net” claim lives, and the public data cannot close it. Only an operator’s audited accounts can.

Capital growth: one index, and it is not for villas

Bank Indonesia’s residential property price survey is the only price index that touches Bali. It covers new landed houses sold by developers in Denpasar, surveyed face to face. For the second quarter of 2026 it shows Denpasar prices up 1.02% year on year, after 0.43% in the first quarter, and primary-market sales down 2.36%, after a 25.67% fall in the first quarter. Nationally, prices rose 0.69%.

That is the index. There is no land price series for Bali from the central bank, the statistics office, Colliers or the listing portals. The “15% to 20% a year” that listings attach to Canggu and Seminyak has no published series behind it; it comes from asking prices on the same portals that carry the listings. Asking prices in the corridor have risen since 2021; no index measures by how much. Whether they go on rising is a forecast, and the one index that exists is running at one percent.

Building costs have a public index too. The statistics office’s construction cost index for 2025 puts Denpasar at 118 against Surabaya at 100 and Gianyar at 106. Turner & Townsend’s survey has Jakarta at USD 943 per square metre and construction inflation around 3% a year. For off-plan buyers the “20% to 25% appreciation during the build” that developers quote is the difference between their pre-sale price and their completed price, set by the same developer. It is a discount for taking construction risk, named as a gain.

The lease is a wasting asset

Most of what foreigners buy in Bali is a lease of 25 to 30 years. Horwath and C9 found freehold at 23% of branded residence supply in 2026, up from 12% a year earlier, which is the one published sign that buyers have started to price the term and developers have noticed. The ownership routes are in Can foreigners buy property in Bali? and the comparison in leasehold vs freehold in Bali. What belongs here is the arithmetic.

A 25-year lease bought for USD 298,000 loses one twenty-fifth of its term every year. Straight-line, that is USD 11,920 a year, or 4% of the price. After five years the asset is a 20-year lease. After ten, a 15-year one. A buyer in year ten is buying a shorter right with the same roof, and the extension that would restore the term is priced by the landowner at the time, under a clause that may or may not bind. What happens when a Bali lease expires covers the clause.

Set the 4% against the yield. On the published figures a short-let villa clears about 5.8% after tax and before any running cost, and about 4.3% once a 20% operator fee is out, with staff, power and furniture still to pay. The term consumption is about the same size as what is left. Whether the resale market lets a seller recover that, by paying for a 20-year lease what it paid for a 25-year one, is the question nobody has data on. REID, Colliers and the academic literature publish nothing on how Bali buyers discount remaining term, and days on market for leasehold resales are not reported anywhere we could find. The second-hand leasehold market is unmeasured, and that is the whole statement.

The alternative, dated

A yield means little without the rate it has to beat. In September 2026 Indonesia’s ten-year government bond paid 7.09%, with the policy rate held at 5.75%. The United States ten-year reached 5.27% on 6 October 2026, a level last seen in 2007. A Bali villa’s gross yield of 7% to 8.3%, before costs, before term consumption, before a flood, sits beside a sovereign bond at 7.1% that requires no operator.

Currency cuts the other way for a dollar buyer. The rupiah went from 14,219 to the dollar in December 2021 to 17,618 in September 2026, a fall of 19%. A villa bought in rupiah and sold in rupiah lost a fifth of its dollar value since December 2021 before any change in the local price. Short-let revenue is often priced in dollars and hedges part of that; a long let to a local tenant, in rupiah, does not.

Supply, licences and the sealed villas

The ministry of tourism said in January 2026 that more than 29,000 non-hotel units in Bali were listed on booking platforms and about 14,500 of them held a business licence in the national registry. Half the short-let stock, by that count, was unlicensed. In February the minister gave operators until 31 March to register; in May the deadline moved to 31 May. No regulation number was attached to either date, and the pages that call 31 March 2026 a “compliance deadline” for construction have the wrong instrument. It concerned listings. Buildings were never in it.

Enforcement on the ground has been about land. On 30 December 2025 the provincial assembly’s special committee on spatial planning, with Badung’s civil enforcement unit, closed 30 villas in Babakan Canggu built on protected rice fields without building approval. In June 2025, 45 villas and restaurants at Bingin were flagged on coastal land. In September 2026 seven villa owners on protected paddy in Subak Munggu, Mengwi, were summoned. The provincial regulation on productive land and nominee ownership, Perda Bali 4 of 2026, signed on 24 February 2026, gives the sanctions a ladder from written warning to demolition and names the people who arrange nominee deals alongside the parties. Six regencies agreed in July 2025 to issue no new hotel or restaurant permits from 2026.

What none of this produces is a count. The nearest thing is Badung’s own 2024 survey, which found 226 villas registered as private homes and put them on the tax roll. There is no published total of villas sealed in 2026, no total of pondok wisata licences in Badung or Gianyar, and no published case under the nominee article yet. The direction is clear. The scale is anyone’s guess.

The risk register, with dates

None of these appears in a yield table. All of them are in the newspapers, with numbers, and an owner’s insurer reads the same newspapers.

What nobody publishes

The list is short and it is the list that matters.

• A land price index for Bali, or any area of it.
• Villa occupancy by area with a stated method. The three services disagree by 27 points for the island; by area they are samples of a few thousand listings, 3,920 for Canggu in AirROI’s case.
• A survey of what villa management costs. Every figure in circulation is a price list.
• How buyers price remaining lease term, or how long leasehold resales take to sell.
• The number of villa licences in Badung and Gianyar, and the number of villas sealed in 2026.
• Any audited net return on a villa portfolio in Bali, from anyone.

A yield claim that fills one of these gaps with a precise number has made the number up, or taken it from a single portfolio with the method undisclosed. Treat it as a price, not a measurement.

How to read a yield claim

Six questions sort the serious listings from the rest, and a seller who can answer them is worth talking to.

1. Gross or net? If net, net of what, listed.
2. Which occupancy, from which source, for which year? “Our data” with no method is a sales figure.
3. Which nightly rate, in which season? The AirDNA island average is USD 130; a listing assuming USD 300 is assuming Seminyak’s average (USD 278 in AirROI) for the whole island, all year.
4. What is the price basis: how many years of lease, and is the extension priced or merely mentioned?
5. Which taxes are inside the figure: the 10% on the lease payment, the 10% hotel tax, the 20% withholding for a non-resident?
6. What exit does the figure assume? A payback period that ends at year six says nothing about year seven.

With those answered, the listings on Bali villas for sale and Bali apartments for sale can be compared on the same basis. Which bedroom count and which product earn what, on third-party data, is in 1, 2 or 3 bedrooms and villa or apartment in Bali; where the permits exist is in where to buy property in Bali.

Bali property investment: FAQ

Is 2026 too late to invest in Bali?

The question assumes a cycle with a top, and no index exists to show where Bali is on it. What the data does show is a market whose demand went flat in 2026 while 5,641 hotel rooms and more than 70 residence projects are in the pipeline. That is a market that rewards buying well and punishes buying on a brochure. It has not closed.

Which earns more, a villa or an apartment?

On Global Property Guide’s long-let figures the two- and three-bedroom units yield most, 5.5% and 5.2%, and the largest the least, 2.9%, because price rises faster than rent with size; the one-bedroom, at 3.3%, is dragged down by its asking price. The short-let data by bedroom is in the bedrooms article linked above. The structural difference is simpler. A villa let by the night is a hotel business with staff; an apartment let by the month is a tenancy.

Can a foreigner get a mortgage in Bali?

In practice, no. Bank Indonesia said in 2016 that no regulation prevents a bank from lending to a foreigner, and in 2025 one bank offered a product to stay-permit holders for a single Jakarta project. For a Bali villa on a lease there is no collateral a bank will take, and purchases are cash. The return has to be judged unleveraged.

What return should I underwrite?

A method, since no single number survives the first question. Take a published occupancy and a published rate for the segment, apply the taxes by law, subtract an operator’s quoted fee and a running-cost allowance you have seen on an invoice, then subtract the lease term consumed each year. Compare what remains with 7.1%. If the villa still wins, it wins for reasons that survive the arithmetic, and those are the reasons to buy it.