Two hundred plants per rai

There is a banana grove in central Bangkok, on land worth more per rai than most people will earn in a lifetime, and it is there for tax reasons.

This is entirely legal. It is avoidance, not evasion: the landowner is complying with a published rule, in a way the rule permits, and nothing in this article suggests otherwise. What is interesting is why the rule permits it, because the reason is not carelessness.

The arithmetic

The Land and Building Tax Act B.E. 2562 (2019) taxes land according to what it is used for. The Act itself sets only ceilings — agriculture not above 0.15%, vacant or unused land not above 1.2% — and leaves the actual rates to a royal decree. The rates now in force are in the Royal Decree Prescribing Land and Building Tax Rates B.E. 2564 (2021), which has applied since the 2022 tax year.

Agricultural land: 0.01% on the first ฿75 million, rising in steps to 0.10% above ฿1 billion.

Vacant or unused land: 0.3% on the first ฿50 million, rising in steps to 0.7% above ฿5 billion.

Notice what that second line is not. It is identical to the rate on commercial land. The decree gives ordinary “other use” and “left vacant” exactly the same five bands. In its first three years, leaving a plot empty costs no more than running a shop on it.

The penalty for vacancy is not in the rate. It is in the clock. Section 43 of the Act: land left vacant or unused for three consecutive years is charged, in the fourth year, an extra 0.3 percentage points, and a further 0.3 every three years after that — up to a total of 3%.

So the real comparison is between the agricultural band and the vacant band on the same plot. Take a central Bangkok site with a tax base of ฿500 million, held by a company. Taxed as vacant, the bands come to ฿2.25 million a year. Taxed as agricultural, the same plot comes to ฿215,000. A difference of just over ฿2 million a year, or about ten times over — and that is before the escalation clock starts.

And the price of moving from the one to the other is bananas.

The threshold is published, crop by crop, in a schedule to a joint announcement of the Ministries of Finance and Interior. For กล้วยหอม, กล้วยไข่ and กล้วยน้ำว้า — the three bananas named — it is not fewer than 200 plants per rai.

Two hundred banana plants cost approximately nothing against the sum at stake. That is the whole story, and everything else in this article is about why a well-drafted statute ended up there.

What the law was trying to catch

The Act was aimed at land-banking.

The problem it was written for is real and expensive: valuable central land held vacant for years, generating nothing, while its owner waits for the price to rise. Article 43 Empty by the hundred thousand counted sixty thousand empty condominium units; idle land is the same pathology one layer down. It removes supply from a market that needs it, it makes the city less dense than its infrastructure could support, and it produces no employment, no housing and no tax.

The Act’s stated purposes include improving collection and reducing inequality, and taxing idle holdings harder than productive ones is a reasonable way to pursue both.

So the drafters needed to tax speculation.

And you cannot write speculation into a tax statute.

Speculation is an intention. It is a belief about future prices held in somebody’s head. No assessor can observe it, no form can capture it, and no tribunal could adjudicate it without reading minds. A tax law needs a test that a public official can apply from the pavement, consistently, to thousands of plots, without litigation over motive.

So the drafters used the observable thing instead. Use.

And they wrote the join between the two categories in a way that makes the density schedule do far more work than it looks like it is doing.

The Act leaves “left vacant or not used as befits its condition” to a ministerial regulation, and the regulation of B.E. 2562 (2019) defines it. Clause 2(1) is the hinge. Land that by its condition could be farmed, but whose use over the past year did not meet the criteria the two ministries announce, is not merely outside the agricultural band. It is in the vacant band.

So the plant count is not a discount. It is the border. Two hundred banana plants per rai is the line between the cheapest classification in the statute and the most expensive, with nothing in between.

The regulation also gave the assessor judgement, and it is worth noticing because it goes unused. Clause 4 says that in deciding whether land is vacant, regard shall be had to the environment, the topography, the soil, the slope of the ground, and the use of neighbouring land. That is a test about whether a plot is plausibly a farm in its context — exactly the question the schedule cannot ask. A plant count is auditable and a judgement is arguable, and this site’s reading is that the count is what gets used.

And clause 3 excludes some land from the vacant category altogether, including land being prepared for use, buildings under construction, land whose use is restricted by law or court order, and land under litigation about ownership. The top band is narrower than it looks, and nothing in that list involves a plant.

Which is a good proxy right up until somebody notices it is a proxy.

The state reasoning from what it can see

articles 119 and 134 found Thais reasoning from bad signals because no better one existed: the car, because there is no driver record; the horn, because nothing else is legible.

This is the same failure, with the state on the reasoning end.

The state cannot see intent. It can see plants. So it taxes plants, and it gets plants.

The land has not changed hands, changed purpose, or entered productive use in any sense its owner cares about. It is still being held for capital appreciation, by the same person, for the same reason, on the same timetable. What changed is that it now satisfies a test.

And here is the design principle underneath, which is worth stating on its own:

A proxy only works if it is expensive to fake.

Article 80 Money you can wear found that gold’s 96.5% purity standard protects the least sophisticated buyer in the market. It works because you cannot cheaply fake purity — the test is real, the cost of defeating it exceeds the gain, so the standard holds.

“Agricultural use” was fakeable for the cost of a few hundred banana shoots against a tax bill in the millions. The proxy cost less to satisfy than to avoid. A test that cheap is not a test; it is a formality with a price list.

The densities were not a response. They were there first

This is the part I had wrong, and correcting it makes the case worse rather than better.

It is tempting to tell the story as a state that wrote a loose rule, watched people plant bananas, and tightened it. That is not what happened. The density schedule was published on 30 January 2020, before the first tax bills went out. The number was there from the beginning.

What came later was a recalibration, not a crackdown. A second announcement of 31 January 2023 replaced the schedule outright, and gave its reason: to make the minimum rates per rai more consistent with the facts of agricultural practice. The bananas stayed at 200.

And look at the schedule, because it explains why bananas and not something else. The thresholds are not arbitrary and they are not uniform: mangosteen 16 plants per rai, durian, mango, longan and lychee 20, rubber 80, papaya 100 or 175 depending on the bed, arabica coffee and passion fruit 400. Those are real planting densities. A durian tree needs room; a banana does not.

So the cheapest classification is not the one with the fewest plants. Sixteen mangosteens would clear the bar with a twelfth of the planting. Bananas win on cost and on time — a sucker is close to free, it establishes in a season, and it does not need a decade to look like a plantation. The schedule measures density, which is agronomically correct. The gamer optimises for price per plant and months to compliance, which the schedule does not measure at all.

Specifying a proxy more precisely does not change the behaviour. It changes the compliance.

The land is now held for capital gain with the correct number of banana plants per rai. The intention is untouched. The instrument has become more detailed, the paperwork more exact, and the outcome the same — which is what always happens when you sharpen a proxy rather than replace it.

And now the part that complicates it

I want to give this its due, because a piece that stops at the joke would be missing something real.

Some of it is genuine. Not every plot planted after 2019 is a dodge, some of the produce is actually harvested and sold, and small-scale urban agriculture is not a fiction merely because a tax rate encouraged it.

And even the cynical version has a real externality.

Article 07 Why Bangkok floods — and why it’s mostly not the rain found that Bangkok floods mostly not because of rain but because its drainage was built over — a city partly below sea level, on a slow-falling river basin, with much of its canal network filled in for roads. The single most useful thing an unbuilt plot in that city can do is remain permeable.

A few thousand rai of banana plants, in a city that pours concrete on everything, is accidentally the opposite of paving. It absorbs water. It is cooler than a car park. It is greener than a hoarding.

None of that was anybody’s intention, and it is still true. The tax avoidance is real, the environmental side-effect is real, and an honest account has to hold both.

Which does not rescue the policy. A tax designed to bring land into use has instead produced the cheapest possible simulation of use, and the land is no closer to housing anybody.

Bangkok already has the lever, and it is the wrong one

In July 2026 the BMA said it would raise its land tax rates, and academics at Thammasat’s law faculty published advice on the problem, recommending three things together: rates tied to the city plan, a coherent land-management policy across the system, and a clearer framework for what counts as agricultural land.

The city can do the first part on its own, and the Act is unusually generous about it. Section 37 paragraph six lets any local authority enact a local ordinance charging more than the royal decree, up to the Act’s ceilings — 1.2% on vacant land against the 0.7% now collected, and 0.15% on agriculture against 0.10%. Paragraph seven then requires every local authority to get its rates approved by the provincial land and building tax committee first — and expressly excepts Bangkok.

So the one city where idle land is most valuable is the one authority that can raise the rate without asking anybody. The headroom is real: nearly double on vacancy, sitting unused since 2022.

And raising it does not fix this. A higher vacant rate widens the gap the bananas exist to cross. The steeper the penalty on the category people can leave, the more it is worth leaving — and leaving costs two hundred suckers. Raising the rate makes the avoidance more valuable, not less.

The direction that follows from the diagnosis is different. The instruments that address land-banking elsewhere tax the assessed value of the site regardless of what is growing on it; a site-value approach does not care what you plant, because planting is not what it measures. Tying rates to the city plan, as the Thammasat advice suggests, points the same way — a plot’s zoning is a fact about the land, not about the current season’s shrubs.

That is a large policy argument with serious opponents and this article is not going to settle it. But the diagnosis is narrow enough to state: if use is a cheap proxy for value-holding, a bigger penalty attached to the proxy is not a fix. Changing what is measured is.

Common misconceptions

“It’s tax evasion.” It is avoidance. The landowners are complying with published criteria in a way the criteria permit. Whether it defeats the Act’s purpose is a separate question from whether it is lawful.

“The tax difference is small.” On a ฿500 million central plot the difference between the two classifications is a little over ฿2 million a year, before the escalation clock starts.

“Any planting counts.” A schedule specifies densities crop by crop — 200 per rai for the three named bananas, 20 for durian, 16 for mangosteen, 400 for arabica coffee. Miss the figure and the land is not simply outside the agricultural band; the ministerial regulation puts it in the vacant band.

“It’s only a Bangkok thing.” The rate differential applies nationally; it is most visible in Bangkok because that is where idle land is most valuable.

“Tightening the rules will fix it.” Tightening a proxy produces more precise compliance. The behaviour changes when the thing being measured changes.

“The densities were brought in to stop this.” They were published on 30 January 2020, before the first bills. The 2023 revision states its purpose as making the minimum rates match agricultural practice.

“Raising the vacant rate will fix it.” It widens the gap the planting exists to cross. Bangkok can raise it by local ordinance, to 1.2%, without provincial approval — and doing so makes crossing more valuable, not less.