The quota, not the shortage

Two numbers do most of the work in the restroom-robotics pitch. The first is that restrooms take up about 6 percent of a commercial building's floor area and consume about 60 percent of its cleaning hours. The second is that annual turnover among restroom cleaners runs between 200 and 400 percent. Both appeared in a September 2026 funding announcement [single source], and both are quoted downstream as though they were measurements.

We nearly used both. They are the kind of statistic that makes a market sound urgent without asking anyone to open an instrument.

In Singapore neither number is needed. The case for automating this work is written down in government instruments a reader can open, and none of them is a forecast. The wage floor for a restroom cleaner is a condition of the employer's operating licence. The number of foreign cleaners a firm may employ is capped by statute at 35 percent of its workforce. Past that ceiling, the firm cannot hire another one at any price. That is not a labour shortage. It is a quota, and a quota is a different problem with a different set of answers.

We also traced the two statistics, because an argument built on refusing unsourced numbers owes the reader the same treatment of its own. Neither survives, one is still sitting in our own internal documents, and Singapore publishes a better version of the second.

What the wage floor actually is

Our Cleaning page already charts the Progressive Wage Model floors to 2029, so the schedule takes one line here: the basic monthly wage for an outdoor, healthcare or restroom cleaner is S$2,495 for the wage year from July 2026, S$2,665 from July 2027 and S$2,835 from July 2028 (MOM, cleaning-sector PWM, page last updated 1 September 2026, read 11 September 2026). That is a rise of 13.6 percent in three years, legislated before a single contract for those years is signed [inference].

"Gazetted" undersells the mechanism. NEA's cleaning business licence page, last updated 19 August 2026, makes a progressive wage plan a condition of the licence itself — one "conforming to the wage levels and bonuses specified by the Commissioner for Labour" — under section 80H(1) of the Environmental Public Health Act 1987. Breaching a licence condition attracts penalties to a maximum of S$5,000, and the licence may be suspended or revoked. A cleaning company in Singapore does not choose whether to pay the floor in the way a company elsewhere chooses whether to raise wages. It pays the floor or it stops trading.

The floor is the basic wage and not the cost. On top of it sit a mandatory annual bonus of at least two weeks' basic monthly wages and employer CPF, which is 17 percent for an employee aged 55 and below on rates effective 1 January 2026 (CPF Board, page updated 11 August 2026). Twelve and a half months of wage, loaded at 17 percent, divided across twelve months, is basic × 1.21875. So one restroom cleaner costs an employer S$3,041 a month from July 2026, S$3,248 from July 2027 and S$3,455 from July 2028 [inference].

Two honesties about that number. It is a ceiling, not an average: 17 percent is the maximum employer rate, it steps down with age — 12.5 percent above 60 and up to 65 — and at 12.5 percent the same July 2026 cleaner loads to S$2,924 [inference]. And it applies only to a citizen or permanent resident. A Work Permit holder is not on this ladder, receives no CPF, and has no statutory wage floor at all. That split is the hinge.

• Resident, loaded — Jul 2026 — 3041 S$/month — basic S$2,495 × 1.21875
• Resident, loaded — Jul 2027 — 3248 S$/month — basic S$2,665 × 1.21875
• Resident, loaded — Jul 2028 — 3455 S$/month — +13.6% on Jul 2026, by statute
• Tier 3 levy, one foreign worker — 800 S$/month — 26.3% of the Jul 2026 resident stack
• What one resident cleaner costs, loaded
• What permission for one foreign cleaner costs

That escalator lands on a thin sector. One published cost model — SGAI Studio, Starting a cleaning company in Singapore, decoded, read 11 September 2026 — puts cleaner wages, CPF and the PWM bonus at 58 percent of revenue and net margin at 4 percent [single source, secondary]. We have carried that figure on two pages with the word "reported", and the word is wrong: SGAI's own page introduces the stack as "modelled per S$100 of revenue" and labels the net-margin line an illustrative model. Nobody reported it. One Singapore industry site modelled it, and the mark stays where it is. The same page attributes to SingStat, for SSIC 812 in 2024, operating receipts of about S$3.6 billion across 2,316 establishments, and a gross operating surplus of about 2.4 percent that turned negative in 2022 [single source, secondary]. Those reach us through SGAI too, rather than from SingStat directly, and they point the same way.

The ceiling, which is the part nobody prices

Every firm in this sector runs on a mix of residents and Work Permit holders, and the mix is not the employer's to choose. MOM's services-sector Work Permit requirements, page last updated 1 July 2026, read 11 September 2026: "The DRC for the services sector is 35%." The Dependency Ratio Ceiling caps foreign headcount as a share of the firm's own total workforce. Within that band the monthly levy is tiered by how foreign-dependent the firm already is — S$450 basic-skilled up to a 10 percent share, S$600 above 10 and up to 25 percent, and S$800 above 25 and up to 35 percent. Higher-skilled workers are levied at S$300, S$400 and S$600 across the same tiers.

Read the top tier first, because that is where a firm with a labour-intensive contract sits. The marginal foreign cleaner costs the wage, plus S$800 a month, which is S$9,600 a year of levy per worker before anyone has been paid [inference]. Against the resident stack above, the levy alone is 26.3 percent of what a resident restroom cleaner costs loaded [inference] — for a person the state has already decided you may not employ more of.

Then read the arithmetic of the ceiling, which is where price stops being the variable. If foreign headcount divided by total headcount cannot exceed 0.35, then foreign headcount cannot exceed 0.5385 times local headcount. To add one more Work Permit holder, a firm already at its ceiling must first add 1.857 local employees [inference]. At July 2026 rates that is roughly S$5,647 a month of new local payroll to buy the right to pay an S$800 levy [inference].

The denominator is itself an instrument. MOM's Local Qualifying Salary page, also dated 1 July 2026, counts a local as one full head at S$1,800 gross a month or more, half a head between S$900 and S$1,799, and not at all below S$900. Every rung of the cleaning PWM from July 2026 sits above that line, so every PWM-compliant resident cleaner clears the LQS and counts whole [inference]. The wage instrument and the quota instrument are one coupled constraint, and complying with the first is what makes the second computable.

MOM is explicit about what happens at the wall. A firm over the ceiling may "keep their excess workers until the Work Permits expire" but will "Not be able to apply for new or renew Work Permits in excess of the DRC." No fee clears it.

1. Tier 1 — up to 10% foreignS$450 / month
2. Tier 2 — above 10% to 25%S$600 / month
3. Tier 3 — above 25% to 35%S$800 / month
4. 35% ceiling — cannot hireNo new or renewed Work Permits above the ceiling, at any levy. Existing workers may be kept until their permits expire.

Each step is a band of foreign share of total workforce, and the rate is the monthly levy for one basic-skilled Work Permit holder in that band.

• A levy band: what one more foreign worker costs in that band
• The top band, where a labour-intensive cleaning contract sits
• The ceiling, where price stops being the variable

One more thing about the tiers, which almost everyone reads backwards. From 2028 the current Tiers 1 and 2 merge, at S$600 basic-skilled and S$400 higher-skilled. A firm below 10 percent foreign share does not get consolidation; it gets S$150 a month more per basic-skilled worker, S$1,800 a year [inference]. Tiers 2 and 3 are untouched. The instrument tightens at the bottom, on the firms that depend on foreign labour least.

What we cannot tell you is how close the sector actually sits to its ceiling. About 55,000 cleaners work in Singapore, of whom about 41,200 are residents — 2022, attributed to NEA, reaching us through SGAI [single source, secondary] — but that is a sector aggregate and the DRC binds firm by firm. We have not found the share of this sector's workforce holding Work Permits published anywhere, nor any distribution of how close licensed cleaning firms sit to the 35 percent ceiling. If most firms sit well below it, this is a cost and not a wall. It is the number that would change our mind, and we have not found it.

The two numbers, traced

The 6-percent-of-area, 60-percent-of-hours claim has one located instance: a quote from 薄益群 (Bo Yiqun), founder and chief executive of DayDayUp, in the September 2026 announcement of the Singapore restroom-robot company Hivebotics' funding round [single source]. DayDayUp is named in the same announcement as exclusive financial adviser to the round, so it is an adviser's line in a document he was paid to help produce. Our own landscape note attributes the claim to Hivebotics' founders and cites a Straits Times interview for it; the interview does not contain it, and we have not found them saying it anywhere — not in that interview, not in the English coverage of the round, not on the company's own site, which publishes no such ratio. Nor have we found an independent measurement of it, in any market.

Singapore publishes primary substitutes that do the same job. The Restroom Association (Singapore)'s A Guide to Better Public Toilet Design and Maintenance, 5th edition 2022, publishes a productivity indicator for washroom cleaning, with its calculator hosted on NEA's own productivity-indicators page: a 25-square-metre washroom with three toilet bowls and four urinal bowls takes a WSQ-trained attendant 39 minutes and 13 seconds. Annex A Table 1 of the National Cleaning Standards for Acute Healthcare Facilities 2024, which the National Infection Prevention and Control Committee wrote on the Ministry of Health's commission, sets 10 minutes once a day for a single-room toilet, 30 minutes for a toilet with a shower, and 45 minutes for a common toilet. A ratio tells you a room is disproportionate. A norm tells you what the work costs in minutes, dated and published by the bodies whose standard the work is judged against.

The turnover figure is worse, and the provenance is the finding. Rapid Eye Inspections' review of cleaning-industry turnover statistics, published 1 August 2026, went looking for the primary behind "averaging around 200% and sometimes reaching as high as 400% annually" and reports that the only named document is a 2024 janitorial-industry whitepaper from 4M Building Solutions. Versions of the figure "repeat across the industry without a named primary source anywhere in the chain," it says, and "we searched for one and could not find it" [secondary]. The funding announcement credits the range to Kimberly-Clark, and Kimberly-Clark Professional does publish it: a page on retaining janitorial staff, read in the Internet Archive's 15 August 2025 capture, has the industry "experiencing a staggering 200-400% turnover rate during normal times" [single source]. Its footnote points at an IBISWorld janitorial-services industry report of October 2024 — a subscription market-research product, not a published benchmark. So the chain is three deep and ends where a reader cannot follow it: a hygiene-products manufacturer's marketing page citing a paywalled industry report, a contractor's whitepaper asserting the same range with nothing named behind it, and a reviewer who chased both and found no primary at the end of either.

The nearest thing to a benchmark is old, small, and reaches us at one remove. In the ISSA and BSCAI 2017 building-service-contractor survey, as reported in that same review, 41 percent of firms reported annual turnover of 50 percent or more and 20 percent reported under 10 percent [secondary]. That is 64 qualified firms at a 4 percent response rate, a sample worth disclosing in the same breath as the percentages.

And then there is the number nobody in this category quotes, which is free, Singaporean and current. MOM's labour-turnover series, obtained from SingStat TableBuilder table M184051, data last updated 20 March 2026, puts the average monthly resignation rate for cleaners and labourers within Cleaning and Landscaping at 2.2 percent in 2025 and 2.6 percent in 2024. Against an all-industry average monthly rate of 1.2 percent in 2025, cleaning runs about 1.8 times the national rate [inference] — a real and quotable fact about a hard job. Annualised arithmetically, 2.2 percent a month is about 26 percent a year [inference], an order of magnitude below 200 percent. Two disclosures come with it: the annualisation is ours and ignores churn within a month, and the series covers MOM's surveyed establishments, so whether it reaches small cleaning firms and Work Permit holders is not established.

We are not clean on this. The 200-to-400-percent range is still the labour frame in two of our own internal documents, written in September 2026, after our landscape read had already flagged the figure as soft. Publishing that is the point. A method that only audits other people's numbers is not a method.

What an automation case looks like without a forecast

Strip out the two statistics and the case is unchanged, which is the argument. A restroom cleaner's statutory floor rises 13.6 percent in three years, enforced against the employer's licence rather than its conscience. The loaded cost of that cleaner reaches S$3,455 a month by July 2028 [inference], against a sector one published model puts at 4 percent net margin [single source, secondary]. And the alternative — employ a foreign cleaner instead — is capped at 35 percent of the workforce, priced at S$800 a month at the top tier, and unavailable beyond it, at which point the firm's only route to one more pair of hands is 1.857 more local ones [inference].

For an investor who has heard the labour-shortage pitch in five markets, the difference here is that nothing in the case above is a projection. Every statutory figure comes off an agency's live page with a date on it, the one modelled figure is marked as a model, and the one number we would most like — how close firms actually sit to the ceiling — is missing rather than estimated. For an FM contractor or a hospital facilities manager, this arithmetic is already on your payroll, and the next step depends on your headcount mix and your contract term, so you can do it better than we can.

We have not done that step here on purpose. The moment a machine's price sits next to a public labour cost, the piece stops being about the instruments and becomes a payback model with a vendor's assumptions inside it. The two statistics this article started with came from a fundraising announcement, a manufacturer's marketing page and a vendor whitepaper. The wage floor, the ceiling, the levy table, the cleaning norms and the turnover series are published by the Singapore government, dated, and free.