The Audit That Passed

On 23 April 2013, cracks appeared in the Rana Plaza building in Savar, outside Dhaka. Workers pointed them out. An engineer looked. The shops and a bank on the lower floors closed. The garment factories above did not, and the workers who raised it were told to go back in.

The next morning the building came down. More than eleven hundred people were killed; the commonly cited figure is 1,134, and sources differ by a handful either way. Around two thousand were injured.

Two of the factories inside had recently cleared social audits commissioned by their buyers.

Those audits were not fraudulent. They were doing the job they were paid to do, which was to check labour practice inside a tenant's operation against a checklist: wages, hours, child labour, fire exits, records. The structural integrity of the building the tenant was renting space in was not on that checklist, because nobody had put it there.

This is the single most useful thing to take from Rana Plaza, and it generalises well past Bangladesh: a clean audit is evidence about one supplier, on one day, against one scope. It is not evidence about a supply chain.

Three gaps do most of the damage.

Scope. The audit checks what the commissioning brand asked about. If structural engineering, electrical load or the legality of the building's upper floors were not in the terms of reference, a perfect report says nothing about them.

Notice. An announced visit measures how a factory looks when it knows you are coming. Undeclared overtime stops, the records are tidy, the fire doors are unlocked for the morning.

Subcontracting. The audited factory is frequently not the factory that sews the garment. Work is passed down to a second or third tier that the brand has never heard of, and no audit of the first tier can see it.

Two schemes were built afterwards, and comparing them is the clearest natural experiment the industry has produced.

The Accord on Fire and Building Safety was signed within weeks, led by largely European retailers including Inditex and H&M. It was a legally binding contract. Independent engineers inspected around sixteen hundred factories, findings were published, and signatories committed money to fund the repairs the inspections required. Binding means a signatory could be held to it.

The Alliance for Bangladesh Worker Safety was the voluntary alternative, joined mostly by North American retailers that would not sign a binding agreement. It inspected on a broadly comparable scale, around a thousand factories, and ran its own programme with its own reporting.

The comparison is not about effort or sincerity. It is about whether a commitment is enforceable, and that is the variable worth watching in any pledge an industry makes about its own supply chain.

If you are the one buying production, the practical questions are short and nobody asks them often enough. What was the certificate scoped to check. Who paid for it. Was the visit announced. Is the factory permitted to subcontract, and if so to whom. Can you name the building.

A brand that cannot name the building its garments are sewn in does not know its supply chain, however thick the document file. The documents are what made Rana Plaza look fine the week before it fell down.