The Compliance Labour Class

I. An Ordinary Day

A reading that should have been taken at 9:47 am was taken at 4:12 pm by the night shift operator instead. The engineer who should have taken it spent the morning on a call she had already answered once before. By the time the reading was finally taken, the pH anomaly had self-corrected. The correction was invisible. The question of why it happened was no longer askable. The numbers that left the factory that evening would look clean.

This is a factory in Faisalabad. It is a Tuesday. The meter sits in a concrete enclosure near the dyehouse. Under normal circumstances it is read twice a shift by the person whose judgement was formed over seven years of watching what water does when something upstream changes. That person was at her desk answering the call. The call had a deadline. The meter did not.

A worker on Line 14 raised something with her line supervisor during the morning break. A cleaning crew member, a specific manager, something she had not spoken about before. The supervisor said she would pass it on. She mentioned it to the officer it should have reached. The officer said she would come back to it. The rest of the officer’s day was consumed by a Thursday review she had to be ready for. The message was not returned to. The worker did not raise it again.

This is a factory in Dhaka. It is the same Tuesday. The worker is twenty-four. She has never raised anything before. She will not raise anything again for some time.

The secretary in Karachi logged three grievances that week. All resolved within target timeframes. Her dashboard was green. The buyer portal showed the factory as high-performing on worker voice metrics. On the same Tuesday, a worker had approached her to ask whether something could remain between them. She had explained that the system required formal lodgement for any grievance to appear in the record. The worker listened. The worker did not lodge. The dashboard stayed green.

These are not three bad days. They are three ordinary ones.

II. The people who now carry proof

The engineer in Faisalabad trained in chemical engineering. She runs wastewater, chemical inventory, heat stress protocols, and fire safety across two units. In 2018 her days were organised around five floor walks. In 2026 they are organised around eleven dashboards and a phone that does not stop.

Her laptop opens before she enters the unit. A Higg FEM evidence request is waiting, flagged because a data point she submitted three weeks ago needs clarification. She answers it. At nine she should be on the floor. She is on a video call with a verifier confirming numbers she has already verified. At eleven-thirty she walks the unit with her phone held slightly forward, photographing needle guards and fire extinguishers for the monthly evidence pack a different buyer requires. She greets the line supervisors. She does not stop. At two she drafts responses to an SLCP Corrective Action Plan whose CAPs have been open six months because they require capital expenditure her managing director has not yet approved. At five-thirty the walk she actually wanted to take, the one that would have caught the air leak forty-eight hours earlier, moves to tomorrow. At nine that night she finishes CAP responses at her kitchen table while her children sleep in the next room.

Her factory has measurably better documentation than it did in 2018. Whether its wastewater efficiency has improved in the same period is harder to say. The documentation is easier to measure than the water.

The officer in Dhaka has a business degree. Her title is Social Compliance Officer. In 2016 the job meant preparing the factory twice a year for scheduled audits. In 2026 it means managing a continuous stream of data into seven buyer portals and three assessment frameworks at once.

She opens the first portal. A verifier from last month’s SLCP assessment has posted a follow-up query. She answers. Before she finishes, a different buyer requests worker participation committee minutes from a quarter two years ago. She searches. She finds. She sends. Her line supervisor knocks. The worker’s message. She nods. She will come back to it. A portal notification: a CAP deadline has moved forward three days. She opens it. She drafts. Her phone buzzes. A third buyer wants confirmation that last quarter’s grievance rate was calculated according to their methodology, not the one she used. She recalculates. She submits. The supervisor does not knock again. The worker’s message is not lodged, because it was never lodged. It exists only as something the supervisor remembered and the officer almost did.

Of the audits her factory completed this year, no finding led to a change she had not already recommended to management eighteen months earlier. She has stopped counting how many times she has submitted the same data to different buyers in different formats. It is the counting itself that exhausts her.

The secretary in Karachi is formally an HR assistant. Grievance redressal is one of her responsibilities. In 2017 it involved a physical logbook reviewed quarterly. In 2026 it involves a digital platform with categorisation codes, MIS reporting, and monthly trend analyses routed upward to senior management and outward to two buyer portals.

She enters complaints into the system using its seven-category taxonomy. Two this week do not fit. She files them under Other. Other does not appear in the monthly trend report. A worker approaches her to raise something informally. She explains what the system requires. The worker listens. The worker does not lodge. The dashboard stays green. The buyer portal shows the factory as high-performing.

She knows, in an accounting she does not share, of at least eleven situations this quarter that would meet her own understanding of what a grievance is and do not appear in her system. She files the monthly report. The system’s definitions are what exist.

Role Mutation, 2015 to 2026

How three jobs inside the compliance labour class have been bent

Field Taxonomy

Figure 01

EHS Manager

Faisalabad, Pakistan

2018 · Floor Engineer

• ·Five floor walks per day
• ·Wastewater supervision
• ·Chemical inventory checks
• ·Heat stress, fire safety

2026 · Evidence Producer

• ·Eleven dashboards, daily
• ·Higg FEM evidence queries
• ·Verifier calls, recurring
• ·Monthly evidence photo pack
• ·SLCP CAP drafting
• ·Six-month-old open CAPs
• ·Video verifications
• ·Phone held forward
• ·CAP responses at kitchen table, 9pm
• ·Documentation, majority of week

Walks moved to tomorrow.

Figure 02

Social Compliance Officer

Dhaka, Bangladesh

2016 · Audit Escort

• ·Two scheduled audits a year
• ·Document preparation
• ·Auditor accompaniment
• ·Post-audit CAP filing

2026 · Continuous Data Steward

• ·Seven buyer portals, daily
• ·Three assessment frameworks
• ·Continuous SLCP stream
• ·Two-year-old minute retrieval
• ·CAP deadline re-negotiation
• ·Cross-buyer methodology reconciliation
• ·Data re-submission, same numbers
• ·Supervisor’s relay, deferred
• ·Findings already recommended months ago
• ·The counting itself

Messages not returned to.

Figure 03

Grievance Secretary

Karachi, Pakistan

2017 · Clerical Recorder

• ·Physical grievance logbook
• ·Quarterly review
• ·HR assistant contract
• ·No reporting authority

2026 · System Keeper

• ·Digital grievance platform
• ·Seven-category taxonomy
• ·Monthly MIS reports
• ·Management trend analyses
• ·Two buyer portals
• ·Informal approaches declined
• ·“Other” category, quietly filed
• ·Eleven unreported this quarter
• ·Dashboard: green
• ·Contract title: HR Assistant

Complaints not lodged.

III. The category

A workforce has formed inside the workforce. It does not make the garment. It makes the garment legible to people who will never touch it. Its labour is required by the ethical architecture of global trade. Its labour is paid for by the factory alone.

This is the Compliance Labour Class. It is not named in any industry ontology. It does not appear as a distinct line in cost-of-goods calculations. It is not recognised in the Social disclosures of any major reporting framework. It operates across EHS, social compliance, sustainability data, grievance redressal, and internal audit functions. Its defining output is not safer workers or cleaner effluent or higher wages. Its defining output is proof that these things are being attended to.

The modern compliance system did not eliminate labour. It created new labour whose product is legibility.

IV. Compliance work and the compliance of work

In 2019, two safety researchers at Griffith University in Australia, Andrew Rae and David Provan, published a paper in Safety Science that drew a distinction the industrial safety community had been circling for a decade. They called it safety work versus the safety of work. Safety work, as they defined it, is what is performed to demonstrate that safety is being managed: documentation, procedures, evidence trails, audits, sign-offs. The safety of work is what actually prevents harm during day-to-day operations: the inspection, the fix, the judgement call, the conversation with the worker who noticed something off. The paper’s quiet provocation was that these two categories, though related, are not the same. And that the first had begun to grow faster than the second could keep pace with.

Their earlier 2018 paper, Safety Clutter, identified the three mechanisms by which safety work proliferates: duplication, generalisation, and over-specification. The asymmetry is structural. It is far easier to add a form than to remove one. Over time, the volume of safety work grows while the safety of work, constrained by time and attention, does not. Eventually the first consumes the bandwidth the second depended upon. The organisation becomes better at proving it is safe than at being safe.

Compliance Work and the Compliance of Work

After Rae and Provan, Safety Science 2019

Divergence Curve

The apparel and textile sector has reached the same point of divergence, though it has not yet named the condition. Compliance work and the compliance of work are not the same thing. The hours the officer in Dhaka spends assembling an evidence pack are compliance work. The conversation with the worker on Line 14 that she was going to come back to and did not. That was the compliance of work. The CAP screenshot is compliance work. The capital investment in the pump that would have closed the CAP honestly is the compliance of work. The factory that excels at the first does not necessarily excel at the second. In many facilities, the more the first expands, the less the second is practised.

The hours do not vanish. They move. Floor time becomes evidence time. Judgement becomes trace. Engineering attention becomes documentation. An EHS manager with eight hours of daily attention, dividing them between forty percent floor presence and sixty percent documentation, has not lost the forty percent that used to exist. She has converted it. The factory now produces documentation at a rate it cannot produce safety improvements. The documentation is visible to the outside world. The foregone improvements are not.

V. The cost architecture of invisibility

The FOB price at which a garment moves from a factory in Karachi or Dhaka to a port in Rotterdam or Los Angeles is conventionally segmented into four blocks: raw materials, direct production labour calculated in Standard Allowed Minutes, factory overhead, and margin. The direct labour block covers the operator, the cutter, the finisher. Factory overhead covers utilities, rent, depreciation, and administrative staff. Compliance labour, wherever it sits, is absorbed into overhead. Overhead is the first block subject to downward pressure in every price negotiation. The labour of a compliance officer, a grievance secretary, an EHS manager, and a sustainability data analyst is therefore not merely uncompensated as a distinct category. It is located in the part of the cost architecture most vulnerable to erosion.

These costs are not invisible because they are small. They are invisible because they are absorbed into roles the factory is already expected to make endlessly elastic.

Where the Compliance Officer’s Hour Lives in the Price

Two FOB architectures. One structural invisibility.

Cost Architecture

Conventional FOB

Standard Model

Raw materials

Direct labour

Factory overhead

Margin

Overhead contains: utilities, rent, depreciation, administrative staff

+ compliance labour, absorbed and unnamed

ACT Reform, 2024

ACT Ring-Fenced

Cambodia, 2024 onward

Raw materials

Direct labour

ring-fenced

Factory overhead

Margin

Overhead contains: utilities, rent, depreciation, administrative staff

+ compliance labour, still absorbed

In 2018 the public policy scholars Pamela Herd and Donald Moynihan published Administrative Burden: Policymaking by Other Means, identifying three categories of cost borne by citizens navigating complex state systems: learning costs, compliance costs, psychological costs. Their framework was developed in the context of citizen-state interactions. It extends, with some care, to the buyer-supplier relationship. The supplier bears learning costs in mastering each new reporting framework. The supplier bears compliance costs in the time and labour required to produce evidence. The supplier bears psychological costs in the chronic low-grade anxiety of a buyer relationship whose continuity depends on the unblemished performance of a function that was never funded.

Within this labour class, certain roles have become disproportionately feminised. Grievance committee secretaries, social welfare officers, sustainability data entry staff, worker engagement coordinators. Across Bangladesh and much of Southeast Asia, more likely to be women. The technical engineering roles that command authority in audit-response settings, fire safety engineering, chemical management leadership, EHS director positions, remain more likely to be held by men. The administrative tier, where documentation volume is highest, tends to be paid less than the technical tier, where visibility and authority are higher. The compliance architecture carries a gendered asymmetry in which the most quantitatively burdened roles hold the least structural power. Pakistan, where the textile workforce is less feminised overall, complicates this pattern without erasing it. It is a layer worth naming. It is not the argument of this article.

In 2015, a coalition of global brands, retailers, and the IndustriALL Global Union formed the Action, Collaboration, Transformation initiative, whose Labour Costing Protocol requires that the direct labour component of the FOB price be ring-fenced, separated from overhead, and protected from downward negotiation. In May 2024, the architecture produced the first brand-supported collective bargaining agreements in Cambodia, binding signatories to maintain sourcing volumes and protect wage floors. The mechanism works. It has not been extended to compliance labour. The sewing operator’s minute is now identified and defended as a discrete cost block. The compliance officer’s hour is not. The architecture for costing one exists. The architecture for costing the other does not.

The Social and Labor Convergence Program, ten years after its founding, operates across more than seventeen thousand facilities in over a hundred and twenty countries. Its own 2026 retrospective reports that the average of its assessments was shared 2.96 times in 2024, unlocking an estimated thirty-nine million dollars in audit costs that would otherwise have been duplicated. The figure is real. It is also roughly twenty-nine percent of what SLCP itself estimates the convergence ceiling could eventually unlock. The remaining seventy-one percent is trapped in an ecosystem of non-interoperable frameworks whose labour is absorbed by the suppliers navigating between them.

VI. The curve already walked

In 2016 a research team led by Christine Sinsky of the American Medical Association published a time-and-motion study in Annals of Internal Medicine that tracked how American physicians actually spent their working hours. Across four specialties in four states, physicians spent 27% of their total time on direct clinical face time with patients and 49.2% on electronic health record and desk work. For every hour of direct clinical face time during the working day, physicians spent nearly two additional hours on documentation. One to two more hours of documentation were completed at home after working hours. The researchers did not coin a term for this. The profession coined one for them. It is called pajama time.

The resemblance is structural. A professional class whose training and authority were calibrated for direct engagement with the object of their expertise finds its time progressively consumed by the documentation of that engagement. The documentation serves stakeholders who are not present in the room. The documentation generates data the professional rarely revisits. The documentation grows faster than the underlying practice can absorb it. The profession exhausts itself. A secondary labour category begins to emerge to handle the overflow. In American healthcare it was the medical scribe, a profession whose entire function is the translation of clinical encounter into acceptable record.

In American medicine, institutional recognition arrived a decade after the pattern itself. The 2017 CMS initiative called Patients Over Paperwork was the first regulatory correction. It is still underway.

The global apparel and textile sector has not reached the point of recognition. It has reached the point of pattern.

VII. Performative Ethics

The regulatory architecture of ethical trade has, in the last decade, delivered levels of visibility the global supply chain has never before possessed. Brands can see the chemical inputs of a dyehouse, the hours worked on a sewing line, the grievance rates of a worker population, the emissions intensity of a unit of production, the social compliance score of a Tier 2 supplier in a country the brand has never visited. The system has become capable of extraordinary sight. It has developed an ethical imagination narrower than its evidentiary appetite.

It does not see the person producing the visibility. It does not see the hours that have been converted into evidence. It does not see the conversations that did not happen, the floors that were not walked, the readings that were missed, the grievances that were not lodged because the system that exists to receive them requires a form the worker cannot read. It does not cost the labour of legibility. It does not compensate it. It does not name the class of workers whose daily labour now sustains the architecture of ethical trade.

The factory did not become more ethical. It became more documentable.

If the moral integrity of global trade now rests on a workforce structurally necessary and structurally unseen, whose hours produce the proof the system relies upon and whose conditions the system has no framework to account for, then the question is not whether the system is ethical. The question is whether the system knows what it is looking at when it says it can see.