Journal · 2026.07 · 20 min

How Greenhouse Costs Are Calculated in Taiwan — Subsidy Rates, Published Unit Prices and Full Contract Prices | 2026 Update

A subsidy baseline, a published figure and a full contract price are three different things. This piece explains what each can answer and how to read it, works through one 2015 dataset to show price-index conversion and its limits, and closes with a 17-point self-check list.

First, the figures you can find online, side by side with the levels Hwa-Nan actually uses for preliminary planning.

Please note: cost structures, scope of service, site conditions and project requirements differ between companies. These figures are for preliminary budgeting only and should not be used to compare, judge or rank other contractors’ quotations.

The table is a reference for rough orders of magnitude, but it cannot be used to set a budget — because almost none of the left-column figures state three things: which year the price is from, what grade of facility was sampled, and how far the scope of work extends.

“What does a greenhouse cost per ping?” It is the first question in almost every initial meeting. From a few thousand NT dollars per ping for a flat-trellis net house, to structural greenhouses starting in the tens of thousands of NT dollars per ping, with foundations and MEP priced separately, to research facilities where equipment costs more than the structure. The spread exceeds a factor of ten.

That is not a chaotic market. Under the single word “greenhouse” sit several kinds of engineering with entirely different purposes. One quote may cover only steel and film; another may include fixed foundations, drainage, utility connections, climate control, design, construction management and warranty. Both can be written as “steel greenhouse, lump sum” — and mean completely different things.

So the question is not only “how much per ping”, but:

“What design, what quantities, what site conditions and what operational requirements was this price calculated from?”

Hwa-Nan reads a greenhouse the way it reads a building: requirements before form, drawings before quantities, quantities before price. What follows is the estimating method I end up explaining in every first meeting. We are writing it down so that you have your own framework for judgement before you build. As for construction without drawings — and how those prices are arrived at — we genuinely have no idea. We cannot read minds; a price with no drawings behind it is nonsense.

It explains the three things most often confused online: what the subsidy baseline can answer, how published cost figures should be read, and how a formal budget is actually built from requirements, site conditions, drawings, specifications, quantities and current-period unit prices. Along the way we work through one real published dataset to demonstrate how to shift historical costs forward with a price index — including where that method stops being valid.

This is a long read. If you only want to confirm the direction you are heading in, these four points are enough; use the contents below to jump to whatever you want in detail.

This is the single most common error in comparing quotes — and the least often questioned.

First, the unit subsidies are announced in. Caps for greenhouse facilities are published per hectare; for readability this article converts them to per 0.1 ha (1,000 m², about 302.5 ping). Eligible area is in principle counted in hectares, truncated to two decimal places, and always subject to the authority’s determination. Note also that the traditional Taiwanese fen (分地) is about 293.4 ping — some 3.1% smaller than 0.1 ha. On a multi-million-dollar job that gap is already material.

Second, which area the quote is using. One greenhouse has at least three areas, and they differ substantially:

The most common budgeting error is not arithmetic — it is mixing four different kinds of money into one figure. Construction cost, subsidy, construction-period cash flow and long-run running cost are separate questions. Calculate at least these four:

Each answers a different question: can you afford to build it; what do you pay after subsidy; will you run out of cash mid-build; and what will owning it for fifteen years cost. Adding them together is where most budgets go wrong. A construction-period funding gap is a timing difference between payments and receipts, not an extra cost; a maintenance reserve is future spending and does not belong on top of the first construction contract.

Under the Agriculture and Food Agency’s 2026 Smart Disaster-Resilient Facility Agriculture Program, facility subsidy is 50% of eligible cost in western Taiwan, 60% in the east and indigenous areas and 65% on outlying islands; climate-control and production equipment is subsidised at 50% in principle. Taking western Taiwan (figures converted here from the per-hectare published values):

That right-hand column is where most of the “market rate” and “minimum cost” figures online come from — the per-ping greenhouse cost summaries on outsourcing quotation platforms and agricultural information sites are, for the most part, calculated exactly this way.

Dividing the cap by the rate is not a calculation error — the error is the meaning attached to the result.

For a structural steel greenhouse, a NT$900,000 cap per 0.1 ha divided by 50% gives NT$1.8M. The correct name for that NT$1.8M is:

“the level of spending at which the subsidy cap is exactly exhausted.”

That is: once eligible spending reaches NT$1.8M, the 50% calculation hits the NT$900,000 ceiling. Spend more and that subsidy does not grow.

One misreading needs clearing up: half of that NT$1.8M was always the owner’s own money. So the problem is not that the figure “ignores your contribution” — it is three more practical things:

• It is only the point at which the ceiling binds, not an actual cost. Beyond it, every additional dollar is 100% yours.
• Eligibility itself excludes things. Site levelling, external water supply, power feeders, non-subsidised equipment, design and administration are often not eligible items. That money is still spent — it just never appears in the back-calculated figure.
• It defines no specification and no scope. It says nothing about height, span, foundation type, structural performance, material grade, corrosion protection, covering, works included or warranty. It is just an amount.

Deals close near that number in the real market: a lower price can come from larger scale, favourable site access, higher standardisation, in-house crews or a narrower scope — or from lower material and performance specifications. So the right conclusion is not “can it be built for this”, but: it guarantees no particular specification or scope. Before using it as a budget, answer what kind of greenhouse it corresponds to.

Take a 0.3 ha structural steel greenhouse in western Taiwan. After design and estimating, eligible cost under this item is NT$24M (shell, covering and foundations) — a fully equipped case well above the per-ping starting level quoted earlier; the figure is here to demonstrate the subsidy arithmetic.

And it does not end there. Suppose the site also needs NT$800,000 of levelling, special drainage, external water supply and a power feeder — none of it eligible under this item:

On cash flow, return to the third of the four numbers. Approval is not receipt, and construction payments usually follow contract milestones — so what you need to estimate is not “NT$22.1M sitting in hand up front” but the peak funding requirement: put payments due, subsidy receipts and loan drawdowns on one cash-flow schedule and find the widest gap. Financing changes where the money comes from and when it is paid; it does not reduce the cost of the work.

The reverse also holds: subsidy is not topped up to the cap. For a 0.1 ha simple film house with NT$800,000 of approved eligible cost, 50% gives NT$400,000 — below the NT$450,000 cap, so the subsidy is NT$400,000. It is not rounded up.

Search for greenhouse costs in Chinese and you will land on NT$2,500–3,500 per ping for simple houses and NT$4,000–5,000 for structural ones. Most of it is years old. After a cumulative 30–40% rise in construction prices, budgeting from those figures leaves an obvious gap.

Before offering our own indicative levels, we should be explicit about what they are.

They have exactly one use: to tell you whether you are preparing to spend three million or well over ten.

This is deliberate: the two products have fundamentally different price structures. A simple film house exists to buy basic rain and insect protection at the lowest cost. Its configuration is narrow, both ends of the range are knowable, and the band is honest — price is the product’s whole advantage.

A structural greenhouse is a platform carrying a production system. Nobody invests in one just to stretch film over it — it almost always comes with MEP and climate control, and upgrading the covering to polycarbonate or glass, or the controls toward precision, is simply the same structural logic taken further (a glass greenhouse is a structural greenhouse). There is no honest upper bound to write: the ceiling is set by what you want the greenhouse to carry. Forcing one produces two outcomes only — fully specified projects look like overruns, or the stated ceiling is so high it frightens everyone away.

So we publish only a starting point with defined content: from about NT$15,000 per ping, covering a structurally designed, code-compliant wind-rated shell with construction drawings, built by a professional contractor under warranty, plus covering (film and insect netting). Everything above that — roll-up ventilation (mechanically driven, an MEP item), pad-and-fan, climate control, supplementary lighting, polycarbonate or glass — is configured item by item against production needs. These are not extras added later; they are options you select.

This deserves its own explanation, because it is one of the largest variables in the whole budget. We would rather put it in plain sight than blend a guess into a unit rate.

A structural greenhouse always has fixed foundations — that is what makes it structural. But foundation cost is almost entirely a function of ground conditions. Isolated footings on competent soil versus fill, soft ground, low-lying land needing raising, or a site requiring piles: concrete volume, reinforcement, excavation depth and difficulty can differ several-fold. The same superstructure on different ground cannot have the same foundation price.

There is only one defensible approach: price foundations against a site investigation. That is why we insist on doing one first — not procedural fussiness, but because without ground data any foundation price is a guess, and guesses eventually appear on your invoice as variations.

“Honestly marking an uncertain item “priced separately” is more responsible than hiding a guess inside a unit rate. What you should watch for is never “this is separate” — it is “this wasn’t mentioned until after we started”.”

In the subsidy classification, “structural steel” describes one thing only: a greenhouse with fixed foundations and steel as its primary structure. It prescribes no member size.

Primary members run from round tube and small-to-medium steel sections up to H-sections and large box sections, matching different spans, eave heights, wind ratings and suspended loads. Scale the steel up and fabrication, connections, foundation sizes and lifting plant scale with it — this is not a matter of buying a few more tonnes. It is why two greenhouses both called “structural” can differ enormously in shell cost.

So when you see “NT$X per ping for a structural steel greenhouse”, ask:

“What section is the primary frame? What span and eave height? What wind rating? What covering material? Does the price include climate control, foundations and external works?”

A far more efficient enquiry than “how much per ping” is one that states the conditions. Prepare:

• Site county and locality (drives wind conditions, transport and permitting)
• Intended area and usable extent
• Crop and growing method
• Preferred greenhouse type and height
• Existing water, power, drainage and road access
• Intended climate-control and production equipment
• Target budget range and schedule

The more complete the brief, the closer any contractor’s number — ours included — will be to reality.

This is a problem specific to agricultural facilities, and worth setting out on its own.

• There is very little source data. Ordinary buildings have public-sector budgeting standards, tender award records, construction statistics and an enormous volume of private transactions for reference. Agricultural greenhouses have no equivalent database; published cost figures come mostly from a handful of surveys by agricultural research and extension stations or academic units.
• Samples are small and skewed to simple types. These surveys often cover only a few dozen holdings, concentrated in the easiest type to sample — low span, no fixed foundations, owner-built or partly owner-built. High-specification projects with full design and construction management are rare to begin with, and rarely enter the sample.
• The priced scope is usually narrower than a real contract. Surveys typically count the facility shell and some pipework, excluding levelling, external utility connections, design, construction management and warranty — all of which are paid for in a real contract.
• The time lag is longer than the cover date suggests. A survey published in 2018 was usually questionnaired earlier, and the facilities in it were built earlier still. The “price year” of the data often precedes the publication year by several years, and the price movement over those years is swallowed whole.

Stack those four and the result is that publicly available per-ping figures generally cannot serve as a contract budget. Nobody is falsifying anything — the data simply has inherent limits.

Taoyuan DARES’s “Economic Analysis of Protected Agriculture: Organic Short-Season Leafy Vegetables in Northern Taiwan” (Research Bulletin no. 83) is still widely cited. Note first: the survey covers simple film greenhouses — greenhouses of different construction cannot meaningfully be compared on cost, so these figures apply only to facilities of the same type. Most articles online describe it as “2018 greenhouse costs” — but 2018 is the publication year. The text itself says:

So the rigorous citation is not “simple greenhouses cost NT$510,000 per fen in 2018”, but:

“a study published in 2018, reporting construction costs compiled from 20 holdings interviewed between August and November 2015.”

The study discloses only that 45% of the greenhouses were built or extended after 2013; it does not list the actual construction and payment year case by case. Which means one thing: these costs cannot be precisely restated to 2026, because we do not know exactly which year each sum was spent.

So what follows is a sensitivity test: provisionally assume all costs share the questionnaire year (2015) as a common base, and use the general construction cost index to observe how far nominal costs may have moved. The result is not a precise restatement of any case, nor a 2026 market rate — it answers only what the same thing would be worth today if prices alone were shifted forward.

Even in a sensitivity test, the base year changes the answer materially. Put the candidates side by side:

Look at the first two rows. Base it on the 2018 publication year and the increase is 34%; base it on 2015, when the survey was actually taken, and it is 40%. Those five to six percentage points are precisely what gets lost when a publication year is mistaken for a price year.

Why not use the four survey months themselves (82.46)? Because that swings to the other extreme. The monthly index for August–November 2015 (82.02–82.93) happens to be a local trough — below that year’s annual average of 83.56 and below every year from 2011 to 2014. Use it as the denominator and the increase rises to 41–45%: a bigger number, but harder to defend, since more than half the sampled greenhouses were built before 2013, when the index stood between 84 and 86, not 82.

Per the DGBAS construction cost index (Taiwan, 2021 = 100), the 2015 annual index is 83.56 and the January–June 2026 average is 116.59 — a cumulative rise of about 40%. Under the assumption that all costs share a 2015 base, multiply the survey figures by that factor:

(The original data is per fen; 1 fen is about 293.4 ping.) Within one survey, highest and lowest differ by a factor of 3.375 — the study attributes this to tube specification, greenhouse height, covering material, internal equipment and site accessibility. It also records that growers mostly used 1/2" or 3/4" galvanised tube to keep costs down, with weaker wind resistance, and only those willing to invest more moved to 1" or 1-1/4". Much of the spread inside a single survey is therefore material and specification — which applies to any comparison of per-ping figures.

Two observations.

First: the 2.5–4 m group averages about NT$3,660 per ping once eleven years of price movement is restored — still about 30% below the lower edge of today’s simple-house range (NT$5,200). That gap is not only inflation. The converted figure comes from a questionnaire that never itemised tax, overhead, profit or warranty, and some responses may include the grower’s own labour; today’s range is a price under full commercial contract. Same number, different contents — that is what “not directly comparable” means.

Second: even the survey’s most expensive case converts to about NT$6,420 per ping — still under half of today’s structural shell starting point. That distance is not inflation; it is specification and contract scope, which is exactly why historic simple houses and modern structural greenhouses are two different grades of product.

The construction cost index has sub-indices. A greenhouse, being mostly steelwork and covering, sits closer to “building works” than to a general index that also carries bridges and roads. Over the comparable window (2018 → H1 2026), the general index rose 34.1% and building works 31.1% — about three percentage points lower, the gap coming mainly from civil works.

Because no 2015 value for the building-works sub-index is available, we make no precise adjustment; the point is directional — using the general index puts our result on the high side, not the low side. That is consistent with the principle of this chapter: not even the conversion method is chosen in our own favour.

Note too that the general index reflects construction input prices overall, not a greenhouse-specific cost index. It suits sensitivity analysis; it cannot assert that any particular greenhouse’s cost moved in the same proportion.

A price index answers one question:

“For the same work at the same specification, considering only changes in material, labour and plant prices, what would it cost today?”

It cannot supply any of the following.

• Scope cannot be confirmed. The study never itemised whether tax, transport, lifting, overhead, profit and warranty were inside the amounts respondents reported, and some may include the grower’s own labour. The converted figures therefore cannot be compared directly with a full commercial contract price — they are neither a market rate nor properly a market floor.
• Specifications differ. The 2015 sample was mostly low, simple structures, some with smaller tube diameters. Building a fixed-foundation, code-wind-designed, corrosion-protected greenhouse on the same plot today is a different specification.
• Conditions differ. Region, haul roads, site accessibility and the diseconomy of small works all sit outside the index.
• It is an index conversion, not a market rate. The ranges in chapter 3 come from Hwa-Nan’s recent projects and enquiries — they are not derived from this table. The two can be cross-checked; neither substitutes for the other.

Summing up the correct use: the sensitivity test yields averages of roughly NT$2,425 and NT$3,660 per ping. Even with eleven years of price movement restored, both sit below today’s full-contract levels — useful for understanding why published data looks cheap, but not citable as a 2026 market rate. Market prices still come from current enquiries, actual specifications and full contract scope.

So the right reaction to a per-ping figure online is not “that’s cheaper than I thought”, but three questions:

“Which year is this data from? What grade of greenhouse was sampled? What scope of work is priced in?”

Press the first one all the way: not “when was the report published” but “when was the money spent”. In the Taoyuan data, publication and survey are three years apart — and the year each greenhouse was actually built is not fully disclosed even by the study. That is why this chapter runs a sensitivity test rather than a precise restatement.

If those three questions cannot be answered, the figure cannot be used to budget.

A reminder in passing: prices are still moving. Between January and June 2026 this index rose from 113.81 to 119.16 — about 4.7% in half a year. As of writing, June 2026 is the latest month published.

A lower price may come from greater scale, favourable site conditions, high standardisation, in-house crews or narrower scope — or from reduced material and performance specifications. A higher one may include fuller design, management and warranty responsibility — or may simply be risk priced in. What matters is never the level itself, but whether the difference can be explained back to drawings, specifications and quantities. Explainable differences are choices; unexplainable ones are risk.

Comparing a greenhouse with a house or an office block and asking which is cheaper per ping points you the wrong way, because the two are spending money on different objectives:

A production greenhouse with a single lightweight covering needs no kitchen, bathroom, interior fit-out, lift or acoustic separation, so its shell can indeed cost less.

That does not mean every greenhouse costs less in total than an ordinary building.

A research or precision-controlled greenhouse may carry close-tolerance air conditioning and dehumidification, supplementary lighting with spectrum control, nutrient recovery, CO₂ dosing, environmental and plant phenotyping instrumentation, backup power, and specialist glass in a high-airtightness envelope. In those projects MEP, controls and research equipment can easily be the largest cost of all, exceeding structure and envelope — the same logic as a high-tech fab: the building is space to carry production, and what is expensive is what sits inside it.

So the useful question is not whether greenhouses are cheaper, but:

“Where is this greenhouse’s centre of cost — structure, covering, or MEP and controls?”

A different centre of cost means a different way to budget and different items worth shopping.

The Public Construction Commission publishes budgeting standards for ordinary building works — offices, classrooms, dormitories and similar. Note what they are: an administrative basis for early-stage cost estimating of public buildings, not market transaction prices.

More notably, the guidance itself says that once a project enters design — when items and quantities are defined — the design budget and tender content should no longer rest on a generic per-area standard but return to specific items and quantities; and building types outside the standard’s scope should be assessed against comparable cases, market conditions and project-specific circumstances.

That is exactly the logic greenhouse estimating should follow — and it is the official document saying so:

“Use per-area estimates early to size the budget; once design is complete, return to items and quantities.”

Building-industry data teaches how cost is broken down and managed; it does not license transplanting a housing rate per ping onto an agricultural greenhouse.

First, a common misconception: it is not true that greenhouses never need a building permit.

Article 8-1 of the Agricultural Development Act creates differentiated paths for agricultural facilities. Broadly, there are two: temporary production facilities on farmland built of specified materials without fixed foundations may be exempt from a building permit; facilities with fixed foundations should in principle first obtain permitted-use approval and then a building permit under the law, with differentiated provisions for small areas and standard drawings — where standard drawings are adopted, parts of the design or construction process may be simplified. Step outside permitted use, or undertake building work, and the building codes apply.

The tiering is reasonable. A 3 m high, 8 m span net house and a large public building both have roofs, but their scale and consequence of failure are nothing alike, and codes have always scaled their demands to use and importance. Apply the full public-building review to every production greenhouse and cost is pushed beyond what most growers can carry — which is not necessarily safer, and may simply mean more people defer replacing ageing structures or turn to something with no design at all.

For an owner, the practical meaning is this: the further your facility sits toward the simplified end, the more drawings, structural calculations and quality control depend on the designer and builder you appointed. Choosing the lightly reviewed route saves time and fees, but moves the gatekeeping into that appointment — which is why knowing who checks your structure matters more than whether a permit exists.

The protagonist of a greenhouse is the crop, not the building. Whether a greenhouse is good is settled by yield, quality, labour efficiency, energy cost and storm losses. Architecture and engineering play a supporting role: let light in, let heat out, get water and power where they are needed, keep the structure standing in a typhoon, and let people and machines get in and move.

A concrete example. “Foliage plants” is not yet a design brief. You still need the species and growth stages, target light levels and shading, temperature, humidity and airflow requirements, bench height and working circulation, irrigation and fertigation method, how people and equipment get in and out, and whether the crop might change later. Only then can height, span, ventilation strategy, covering and equipment layout be decided.

Supporting is not optional. A greenhouse is one of the few building types that must resolve structure, covering, airflow, utilities, controls and crop requirements at once — systems that constrain one another, where optimising any single one can wreck another. Integrating them into buildable, inspectable drawings is where engineering earns its place:

“so that the demands of production do not collide with each other for the first time on site.”

Which is why the order cannot be reversed. Pick the cheapest greenhouse first and then force the crop and equipment into it, and the construction price may be lower — the production result will not necessarily be better.

You cannot infer what one operator can afford from an industry’s gross output. What matters is the individual farm: yield per unit area, price and its volatility, gross margin, channel stability, additional viable cropping cycles, disaster losses avoided by the facility, labour saved, and financing terms. There is one test:

“Affordable annualised facility cost ≤ the incremental operating cash flow the greenhouse produces under a conservative scenario.”

Low-margin, volatile crops generally support a lower build cost per ping. Propagation, floriculture, breeding, research and high-value crops are what can carry higher climate-control and structural performance.

A greenhouse is not an asset you pay for once. This is the fourth number from chapter 1: life-cycle cost.

The covering is the clearest example: film life varies widely with material, thickness, exposure and installation quality, and generally needs replacing within a few years; polycarbonate and glass last far longer but cost much more per replacement. Roll-up motors, drive shafts and vent mechanisms are moving parts and wear out; fans, pads, filters, sensors and controllers each have their own replacement cycles. On coastal or high-humidity sites, add corrosion remediation and hardware replacement.

Only when these enter the annualised cost does it become clear whether the extra paid up front for corrosion protection, material grade and maintainability was worth it.

First, make “centre of cost” concrete. Money sits in different places in different greenhouse types, and so do the items worth shopping:

Proportions can invert entirely between types, and a fixed percentage applied to another type will mislead. Establish where the centre sits, then allocate the budget and your shopping effort accordingly.

Two groups are routinely underestimated. Group 1 is invisible once built, yet decides whether the greenhouse settles, leans, floods or lifts in wind — the same superstructure on competent soil versus fill or low-lying ground cannot carry the same foundation price. Group 5 is often outside the “greenhouse” quote yet indispensable to operating; the NT$800,000 in the 0.3 ha example is exactly this.

As for group 2, cost is not just steel tonnage: at the same 300 ping, a 3 m and a 6 m eave height differ in steel quantity, lateral stability, covering area and construction safety.

In practice a doubled quote usually comes from four places: wind design level (every step up in design wind speed adds steel section, bracing and foundation — “designed to code” and “as much as we can” are two different prices); material and corrosion specification (zinc coating thickness, tube wall thickness, covering grade — invisible, but the price difference is real); foundation depth and type (embedment, concrete volume and reinforcement, which map directly onto wind performance); and quoted scope (whether levelling, drainage, utility connections, climate control and warranty are in).

Whether any of that is visible depends on something more basic: whether there is a design.

Jump straight from site area to total price and everything in between is assumed: column spacing, foundation sizes, steel thickness, gutter capacity, vent area, electrical capacity, even whether a crane can reach the site. The more assumptions, the less reliable the price.

Faced with that uncertainty a contractor has two options: price all the risk in, or quote on the leanest assumptions and handle variations later. Neither is good cost control.

Temporary works, general management and minor sundries are reasonably taken as lump sums. But if foundations, steelwork, covering, drainage, irrigation, power and climate control are all “lump sum”, the owner cannot know how much material is included, at what specification, which item drives the difference, how future additions or omissions will be valued — or whether two contractors are pricing the same thing.

“Without design developed enough to measure, there are no reliable quantities; without reliable quantities, an estimate can only remain a lump sum.”

For a structural steel film greenhouse, steel, galvanising and fabrication are the main cost sources, with extruded aluminium (glazing bars, gutters, vent frames, shade tracks) taking a meaningful share. So movements in steel, zinc and aluminium do feed into quotes — that much holds.

It does not generalise to all greenhouses.

Take glass. Horticultural glass is not architectural glass. Depending on the crop it may involve low-iron high transmission, light diffusion, anti-reflective coating, transmission in specific UV bands (the UV-blocking common in architectural glass can actually harm growth and colouration in some crops), varying haze, or double glazing for thermal reasons.

In practice here, architectural glass is often substituted for horticultural glass to save cost. Short term it saves material money; for some crops it costs growth and quality. That trade should be calculated when the material is selected, not discovered at harvest.

Another shift under way is safety specification: as awareness rises, laminated glass is progressively replacing toughened glass, particularly in roofs. Toughened glass shatters and falls pane; laminated glass holds its fragments on the interlayer and is far less likely to come down in quantity, which protects people working below. That upgrade shows up in covering cost — so when comparing two glasshouse quotes, whether the roof is toughened or laminated is one of the specifications to align first.

If the specification you need is not normally stocked in Taiwan, cost is pushed further by minimum production batches, line changeovers or scheduled production runs, performance testing, overseas freight and exchange rates, breakage spares, and re-procuring small replacement quantities later. In such projects the covering system alone can exceed the steelwork — and watching steel prices is then the wrong instrument.

Design is not about producing prettier drawings. Its function is to find things early: whether columns clash with benches and trolley routes, whether foundation levels are right, whether gutter falls and discharge make sense, whether fans and pads match the intended airflow, whether door widths admit machinery, whether irrigation mains and electrical capacity suffice, whether shade screens, roof vents and roll-up gear collide, and whether there is access to replace film and motors later.

Not every project without full design goes wrong. But one round of demolishing and recasting foundations, one refabrication of steel, one second crane mobilisation or one re-run of a power feeder can cost more than the entire design fee.

“The value of design is that the cheapest revisions happen before construction.”

Japan’s MAFF surveys on greenhouse costs offer a useful observation: about 45% of growers surveyed had experienced schedule extensions, commonly from contractor workload and waiting on materials. In extension cases supplied by some contractors, actual construction cost ran about 2–15% above the original estimate — and most of that was absorbed by the greenhouse contractor rather than charged on to the grower. The same series found that contractors’ estimate line items differ in naming and granularity, making comparison hard for growers, which is why it pushes for standardised estimate items.

Those figures are not a fixed variation rate for Taiwanese projects, nor an equivalent for the risk premium in lump-sum pricing. What they illustrate is this: when information about site conditions, permitting, materials, crews and schedule is incomplete, cost risk does not disappear — it is simply carried by either the owner or the contractor. Defining conditions up front is in both parties’ interest.

The Japanese surveys recommend defining, before enquiring: crop and production period, target yield and budget, area, span and eave height, wind and weather performance, covering and ventilation method, irrigation and climate equipment, ground conditions and haul roads, utilities and drainage, and the contractor’s scope of work. The fuller the brief, the more contractors price on a common basis and the fewer variations arise. An owner can do half of this without any drawings — the seven inputs in chapter 3 are the starting point.

Korea sets design criteria for disaster-resistant greenhouses from regional wind and snow data and issues standard drawings; projects not using them require review by a structural body. The lesson: a standard greenhouse is not one national specification, but a validated standard design applied within defined regional wind conditions, dimensions and use limits.

Applied to Taiwan: coastal sites, open plains, valley wind gaps and hillsides cannot share identical foundations and steelwork merely because the area matches. Wind drives columns, beams, bracing and foundations; heavy rain drives gutters, raising and pumped drainage; ground and groundwater drive foundation sizes; coastal salt drives galvanising, bolts and aluminium; haul roads and lifting space drive transport and schedule. All of it is part of the cost.

It has two stages. And one thing that is fair to everyone: without a design, nobody can tell you how deep the foundations go — us included. Specifications and quantities only have a basis once design is complete and a formal quote is prepared. Not having them at the estimate stage is a function of the stage, not anyone’s failing.

None of these eight requires a drawing — they confirm scope and conditions. Sort them out before enquiring and the answers you get will be far more accurate.

A formal quote need not price every bolt, but the principal foundations, structure, covering, MEP and equipment should trace back to clear drawings, specifications and quantities.

We do not set a per-ping price and multiply by area before knowing the crop, the site, the foundations and the equipment. Estimating runs in three tiers:

We still provide the per-ping or per-hectare conversion. But that number is the result of design and measurement, not the starting point of the estimate. The same discipline applies to how we cite others: with the Taoyuan survey in chapter 4 we did not treat the publication year as the price year, did not pick a base that flattered us, and stated the limits of the test up front — because a cost that can be checked requires that even the figures it cites can be checked.

There is no official per-ping price in Taiwan that fits every site, every crop and every set of equipment. Dividing the cap by the rate tells you when subsidy is exhausted, not where the market floor is. Published figures, limited by sample, scope and vintage, are good for orders of magnitude only — and before using them, establish which year the money was spent and what works it covered. Material prices matter, but which ones depends on whether your dominant cost is steel, glass or equipment.

“A cost that can genuinely be implemented is one formed jointly by defined production requirements, site conditions, drawings, specifications, measured quantities and current-period unit rates.”

A sound greenhouse answers four questions at once: what does the agriculture need; what do the site and hazard conditions demand; what can the cash flow carry; and what do design and measurement add up to? The crop is the protagonist, not the engineering. Its job is to make structure, foundations, covering, drainage, MEP and construction all fit the growing — and to turn vague conditions, item by item, into work that can be calculated, compared and built.

If you are still sizing the investment, prepare the seven inputs from chapter 3 (crop, area, site location and so on) and come and talk. A first meeting can exchange views on scale and direction — and to be clear, complete early planning and design is a paid professional service. Design fees and construction costs are not the same order of magnitude; the cost arising from a single construction error is often enough to pay for the full design. What the fee buys is a basis for every decision that follows.

← More greenhouse insights